|
33660
|
PartSourcer by @yaasshh | Stardance
|
yaasshh
|
1
|
50.47
|
515
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T14:26:23Z", "hours": 37.0, "multiplier": 18.84, "payout": 515, "blessing": "neutral", "text": "PartSourcer Find the cheapest in-stock part for your PCB in one search. Live at “part-sourcer.vercel.app” Search any electronic component and get its stock, price, footprint and datasheet in one place. Every number is stamped with the moment it was read, so you always know how new it is. The useful bit: for a part you searched, it looks for one cheaper drop-in that is actually in stock, and tells you why it counts as a match. Same package, matching specs, real stock, and both prices re-read the same way before any saving gets claimed. If it cannot find a genuine match, it elaborates so. Right now that works for resistors and capacitors. Everything else gets an “no match found”. (Perhaps later implementation) Rules it sticks to: No made-up stock or prices. If the data is stale, you get an error, not a number pretending to be fresh. A part with no published price says “no price”, never $0.00. A zero would read as free and win every comparison. Missing fields are left out rather than filled with a placeholder. Built with FastAPI and React, both on Vercel, with Postgres behind the price history. Part data comes from Mouser in real time, plus an open snapshot of the LCSC catalogue that refreshes about once a day. There is a DigiKey adapter too, and it switches on once credentials are set. Free, open source, no login, no paywall. MIT licensed.", "repo": "https://github.com/yaasshh09/PartSourcer", "demo": "https://part-sourcer.vercel.app"}]...
|
[{"post_id": 84457, "ts": "202 [{"post_id": 84457, "ts": "2026-10-29T17:44:12Z", "hours": 1.2903, "text": "Devlog 7 Since the last devlog I’ve mostly been fixing things, because the free data source I use is still broken. Its search leaves out stock numbers for most filtered results, so I had to stop assuming those numbers exist. The biggest fix was finding a part’s specs. Before, I only looked at the top 100 parts in its package, so some parts never showed up. Now I search for the part’s own value first, and a 10k resistor I couldn’t find before shows up fine. I also made the site honest about missing stock. If the number isn’t listed, it now says “stock not listed” instead of “out of stock”. The cheaper swap finder used to throw those parts out. Now it double checks them before deciding. The browse page also stopped saying “top 100 by stock” when it has no stock numbers to sort by. Smaller fixes: if a BOM row has a blank quantity, I now count its designators instead (R1, R2, R3 is three parts). This works on the site and in the command line tool. I also fixed a bug where a slow currency rate could overwrite the currency you had just picked. I cleaned up the codebase too, removing old notes and leftover wording that read like a robot wrote it, and tidying the site copy. I also wrote my answers for the LCSC API application. Now I’m waiting to hear back."}, {"post_id": 82642, "ts": "2026-10-25T14:22:03Z", "hours": 4.1617, "text": "Devlog 6 Since the last devlog I’ve been working through my list of 60 ideas. Part pages got a lot more useful. Each one now has a price history chart, a hint about when a sold-out part might come back, and whether the part is still being made or is on its way out. Sometimes it also suggests a part with better specs for the same money, and now it can show up to three cheaper swaps instead of just one. Search results also get small badges when a price or stock level changes. I added a currency picker, a watchlist for parts you care about, and RSS feeds that tell you when a part’s price or stock moves. For BOMs, you can now see whether buying everything from one store beats splitting the order between stores. I also made tools for people who don’t want to use the website. There’s a developer page for the API, a command line tool that prices a BOM, a GitHub Action that posts your BOM cost on pull requests, and a KiCad plugin. I tested the plugin on real KiCad boards and found it was missing some part number fields, so I fixed that. Other fixes: the BOM page works better on phones, the colors are easier to read, dark mode looks right, and downloaded BOMs can’t carry sneaky spreadsheet formulas anymore. The free data source I use still has broken search, so I’ve started applying for the official LCSC API. Fingers crossed it gets approved!"}, {"post_id": 79259, "ts": "2026-10-13T07:30:24Z", "hours": 8.02, "text": "#Devlog 5 After the security pass, I mostly worked on new features. I added a Browse section, so you can look through six part categories without knowing a part number. Then I built a compare tray. You pin up to four parts from anywhere on the site and see them side by side. I moved the site to Vercel with a database in Singapore. A small job now runs every night and records prices, so later I can show how a part’s price changes over time. Search had some problems. The free data source I use started returning broken results, so I wrote up what was going wrong and reported it to the people who run it. I also stopped showing results that had nothing to do with what you searched, and I made the “part not found” message more clear. Now it says when a distributor just didn’t answer, instead of saying the part doesn’t exist. You can also sort results by price now. The biggest new thing is BOM check. You paste in your whole parts list, or load a spreadsheet, and it prices every line at the quantity you need. It tells you which lines it skipped and why, and you can share the checked list as a link. Part pages now show each distributor’s bulk pricing. There’s also a quantity box, so you see what you’d actually pay for 250 parts and not just one. If you click a part from your BOM, that box already has your quantity in it. Next, I made a list of 60 more ideas and started building them."}, {"post_id": 51003, "ts": "2026-08-20T14:21:47Z", "hours": 1.0139, "text": "Dev log #5?? Ran a 20-item polish checklist over the site. Ten things were already fine: links all work, buttons all do something, the 404 page is real, errors show up properly. Six items from the checklist were missing, so I built them. Fixed the page titles. Every page was showing the same title because nothing reloads in a single-page app. Now a part page names the part, a search names what you searched. Put a year in the footer, read live so it never goes stale. Dropped the “404 demo” link, that was a leftover dev thing sitting next to real links. Then two actual bugs I only found by measuring in a browser at phone size. The home page really was scrolling sideways: a card grid was asking for 400px inside a 319px space. About had the same thing at 340px. Fixed both. The header was the other one. It needed 390px of room for the logo and five links on a 375px phone. It had been wrapping onto three rows and leaving Contact stranded alone. Now the links tuck behind a menu button below 640px. Header went from 281px tall to 69px. Last thing: the pages were too narrow and floated in empty space. Widened them a lot, 1600px for the main shell and 1400px for the text pages. Width alone would’ve just made long ugly lines, so the card grids reshaped too. FAQ goes two across, the three how-it-works steps share a row, the four-card grids go four across and step down cleanly. Plus a few more minor changes, The project is built enough to be shipped!"}, {"post_id": 50904, "ts": "2026-08-20T11:35:02Z", "hours": 7.1633, "text": "Dev log #3 or #4?? After the site went multi-distributor, I kept going. I added a contact page with my email, Instagram and GitHub, dropped GitHub out of the header, gave the footer real columns with links that work, moved the data source note over to the About page and wrote more there. The front page got more to read too, and the About cards now sit in a clean 2x2 grid. Then I did a proper security pass. I added a middleware that stamps security headers on every API response, including CSP, frame options and HSTS. Wrote a rate limiter that reads the client IP from headers the platform actually writes, never the one a caller can fake, and capped its key table so the limiter itself cannot be turned into the attack. Added input caps: query at 200 characters, page at 50, part key at 128, all refused early so nobody can burn my distributor quota. Added a body size ceiling and tightened CORS. The static site got its own CSP through vercel.json, checked against the real built bundle. I also wrote guard tests: they fail if a stray .env gets committed, if a credential literal shows up in source, if SQL starts joining values instead of using placeholders, or if a Set-Cookie ever appears. Then I documented the whole posture in the backend README, including what I deliberately left out and why."}, {"post_id": 48452, "ts": "2026-08-16T20:03:22Z", "hours": 13.2792, "text": "PartSourcer devlog #2 Started with a simple annoyance: finding a part that is actually in stock means opening six distributor tabs per line of a BOM. Wrote a throwaway script to poke at the free jlcsearch API and see what the data really looked like. Turned out it has no brand and no datasheet. Decided early to show that as blank rather than show wrong results. Also built the backend. FastAPI, a swappable data source so the real LCSC API can drop in later. Also tweaked the frontend a bit. React, inline styles, a loud yellow-and-black look. Then deploy configs, and the site went live on Render. Implemented a price history recorder, so the data is there later even if nothing reads it yet. The big addition. The site finally became multi-distributor: Mouser and DigiKey alongside LCSC. Every part now carries a list of offers, one per distributor, and a per-source status so a partial failure shows as a partial answer rather than an error."}, {"post_id": 27780, "ts": "2026-07-13T17:10:53Z", "hours": 15.7958, "text": "I’ve been working on a personal project called PartSourcer. Essentially, it’s a database, a bit beautified, but a database nevertheless, that scours the internet for major electronic component distributors in bulk, such as Mouser, DigiKey, JLCPCB, and LCSC. It goes through all of these major component databases, fetching real-time data from API keys I generated myself, and outputs the components and their prices. So far, I have the frontend and most of the backend done. The next step is to actually implement the Mouser and DigiKey API keys. I have the Mouser one ready; DigiKey is a bit more confusing to get, but I will get it. I already have the LCSC and JLCPCB integration working for each component, showing real-time stock and unit prices per item. I’m going to keep a very early version of the project open source under the MIT license. Later on, I plan to remove the MIT license and host it on the internet for free, but it won’t remain open source. Only that very early version will stay under the MIT license. Moving forward, my roadmap for the project includes: Completing the actual API implementations. Implementing a side-by-side comparison tool. Adding a large-scale, bulk BOM (bill of materials) upload feature. Creating a price history chart. Implementing a “browse by category” feature. Adding a “biggest savings this week” section to show people where the best deals are and what they can save money on."}]...
|
Title: PartSourcer by @yaasshh | Stardance
Devlog Title: PartSourcer by @yaasshh | Stardance
Devlogs:
I’ve been working on a personal project called PartSourcer. Essentially, it’s a database, a bit beautified, but a database nevertheless, that scours the internet for major electronic component distributors in bulk, such as Mouser, DigiKey, JLCPCB, and LCSC. It goes through all of these major component databases, fetching real-time data from API keys I generated myself, and outputs the components and their prices. So far, I have the frontend and most of the backend done. The next step is to actually implement the Mouser and DigiKey API keys. I have the Mouser one ready; DigiKey is a bit more confusing to get, but I will get it. I already have the LCSC and JLCPCB integration working for each component, showing real-time stock and unit prices per item. I’m going to keep a very early version of the project open source under the MIT license. Later on, I plan to remove the MIT license and host it on the internet for free, but it won’t remain open source. Only that very early version will stay under the MIT license. Moving forward, my roadmap for the project includes: Completing the actual API implementations. Implementing a side-by-side comparison tool. Adding a large-scale, bulk BOM (bill of materials) upload feature. Creating a price history chart. Implementing a “browse by category” feature. Adding a “biggest savings this week” section to show people where the best deals are and what they can save money on.
PartSourcer devlog #2 Started with a simple annoyance: finding a part that is actually in stock means opening six distributor tabs per line of a BOM. Wrote a throwaway script to poke at the free jlcsearch API and see what the data really looked like. Turned out it has no brand and no datasheet. Decided early to show that as blank rather than show wrong results. Also built the backend. FastAPI, a swappable data source so the real LCSC API can drop in later. Also tweaked the frontend a bit. React, inline styles, a loud yellow-and-black look. Then deploy configs, and the site went live on Render. Implemented a price history recorder, so the data is there later even if nothing reads it yet. The big addition. The site finally became multi-distributor: Mouser and DigiKey alongside LCSC. Every part now carries a list of offers, one per distributor, and a per-source status so a partial failure shows as a partial answer rather than an error.
Dev log #3 or #4?? After the site went multi-distributor, I kept going. I added a contact page with my email, Instagram and GitHub, dropped GitHub out of the header, gave the footer real columns with links that work, moved the data source note over to the About page and wrote more there. The front page got more to read too, and the About cards now sit in a clean 2x2 grid. Then I did a proper security pass. I added a middleware that stamps security headers on every API response, including CSP, frame options and HSTS. Wrote a rate limiter that reads the client IP from headers the platform actually writes, never the one a caller can fake, and capped its key table so the limiter itself cannot be turned into the attack. Added input caps: query at 200 characters, page at 50, part key at 128, all refused early so nobody can burn my distributor quota. Added a body size ceiling and tightened CORS. The static site got its own CSP through vercel.json, checked against the real built bundle. I also wrote guard tests: they fail if a stray .env gets committed, if a credential literal shows up in source, if SQL starts joining values instead of using placeholders, or if a Set-Cookie ever appears. Then I documented the whole posture in the backend README, including what I deliberately left out and why.
Dev log #5?? Ran a 20-item polish checklist over the site. Ten things were already fine: links all work, buttons all do something, the 404 page is real, errors show up properly. Six items from the checklist were missing, so I built them. Fixed the page titles. Every page was showing the same title because nothing reloads in a single-page app. Now a part page names the part, a search names what you searched. Put a year in the footer, read live so it never goes stale. Dropped the “404 demo” link, that was a leftover dev thing sitting next to real links. Then two actual bugs I only found by measuring in a browser at phone size. The home page really was scrolling sideways: a card grid was asking for 400px inside a 319px space. About had the same thing at 340px. Fixed both. The header was the other one. It needed 390px of room for the logo and five links on a 375px phone. It had been wrapping onto three rows and leaving Contact stranded alone. Now the links tuck behind a menu button below 640px. Header went from 281px tall to 69px. Last thing: the pages were too narrow and floated in empty space. Widened them a lot, 1600px for the main shell and 1400px for the text pages. Width alone would’ve just made long ugly lines, so the card grids reshaped too. FAQ goes two across, the three how-it-works steps share a row, the four-card grids go four across and step down cleanly. Plus a few more minor changes, The project is built enough to be shipped!
#Devlog 5 After the security pass, I mostly worked on new features. I added a Browse section, so you can look through six part categories without knowing a part number. Then I built a compare tray. You pin up to four parts from anywhere on the site and see them side by side. I moved the site to Vercel with a database in Singapore. A small job now runs every night and records prices, so later I can show how a part’s price changes over time. Search had some problems. The free data source I use started returning broken results, so I wrote up what was going wrong and reported it to the people who run it. I also stopped showing results that had nothing to do with what you searched, and I made the “part not found” message more clear. Now it says when a distributor just didn’t answer, instead of saying the part doesn’t exist. You can also sort results by price now. The biggest new thing is BOM check. You paste in your whole parts list, or load a spreadsheet, and it prices every line at the quantity you need. It tells you which lines it skipped and why, and you can share the checked list as a link. Part pages now show each distributor’s bulk pricing. There’s also a quantity box, so you see what you’d actually pay for 250 parts and not just one. If you click a part from your BOM, that box already has your quantity in it. Next, I made a list of 60 more ideas and started building them.
Devlog 6 Since the last devlog I’ve been working through my list of 60 ideas. Part pages got a lot more useful. Each one now has a price history chart, a hint about when a sold-out part might come back, and whether the part is still being made or is on its way out. Sometimes it also suggests a part with better specs for the same money, and now it can show up to three cheaper swaps instead of just one. Search results also get small badges when a price or stock level changes. I added a currency picker, a watchlist for parts you care about, and RSS feeds that tell you when a part’s price or stock moves. For BOMs, you can now see whether buying everything from one store beats splitting the order between stores. I also made tools for people who don’t want to use the website. There’s a developer page for the API, a command line tool that prices a BOM, a GitHub Action that posts your BOM cost on pull requests, and a KiCad plugin. I tested the plugin on real KiCad boards and found it was missing some part number fields, so I fixed that. Other fixes: the BOM page works better on phones, the colors are easier to read, dark mode looks right, and downloaded BOMs can’t carry sneaky spreadsheet formulas anymore. The free data source I use still has broken search, so I’ve started applying for the official LCSC API. Fingers crossed it gets approved!
Devlog 7 Since the last devlog I’ve mostly been fixing things, because the free data source I use is still broken. Its search leaves out stock numbers for most filtered results, so I had to stop assuming those numbers exist. The biggest fix was finding a part’s specs. Before, I only looked at the top 100 parts in its package, so some parts never showed up. Now I search for the part’s own value first, and a 10k resistor I couldn’t find before shows up fine. I also made the site honest about missing stock. If the number isn’t listed, it now says “stock not listed” instead of “out of stock”. The cheaper swap finder used to throw those parts out. Now it double checks them before deciding. The browse page also stopped saying “top 100 by stock” when it has no stock numbers to sort by. Smaller fixes: if a BOM row has a blank quantity, I now count its designators instead (R1, R2, R3 is three parts). This works on the site and in the command line tool. I also fixed a bug where a slow currency rate could overwrite the currency you had just picked. I cleaned up the codebase too, removing old notes and leftover wording that read like a robot wrote it, and tidying the site copy. I also wrote my answers for the LCSC API application. Now I’m waiting to hear back.
Ships:
PartSourcer Find the cheapest in-stock part for your PCB in one search. Live at “part-sourcer.vercel.app” Search any electronic component and get its stock, price, footprint and datasheet in one place. Every number is stamped with the moment it was read, so you always know how new it is. The useful bit: for a part you searched, it looks for one cheaper drop-in that is actually in stock, and tells you why it counts as a match. Same package, matching specs, real stock, and both prices re-read the same way before any saving gets claimed. If it cannot find a genuine match, it elaborates so. Right now that works for resistors and capacitors. Everything else gets an “no match found”. (Perhaps later implementation) Rules it sticks to: No made-up stock or prices. If the data is stale, you get an error, not a number pretending to be fresh. A part with no published price says “no price”, never $0.00. A zero would read as free and win every comparison. Missing fields are left out rather than filled with a placeholder. Built with FastAPI and React, both on Vercel, with Postgres behind the price history. Part data comes from Mouser in real time, plus an open snapshot of the LCSC catalogue that refreshes about once a day. There is a DigiKey adapter too, and it switches on once credentials are set. Free, open source, no login, no paywall. MIT licensed....
|
1793328724.9135368
|
1793415124.9135368
|
|
41747
|
My own site! by @wxie223 | Stardance
|
wxie223
|
2
|
3.5
|
1
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T14:04:53Z", "hours": 0.0, "multiplier": 1.45, "payout": 1, "blessing": "neutral", "text": "I made a personal site about me! It has 3 pages, a main page, a projects page, and an extracurriculars page. I don’t want this website to just be all about me. I also want to include resources for highschoolers just like me, so I’m planning to add a new page with resources for like SAT, or AP, or any other useful website. Hope you like it! I made this in the personal site workshop so it is still a bit simple.", "repo": "https://github.com/WillyX12/PersonalWebsite", "demo": "https://willyx12.github.io/PersonalWebsite/"}, {"ship_num": 2, "ts": "2026-08-19T19:30:17Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I shipped V1 of my personal site. Its really simple, but I hope it makes ur day.", "repo": "https://github.com/WillyX12/PersonalWebsite", "demo": "https://willyx12.github.io/PersonalWebsite/"}]...
|
[{"post_id": 66322, "ts": "202 [{"post_id": 66322, "ts": "2026-09-12T02:52:38Z", "hours": 0.7303, "text": "Havent been working on it a long time cuz of school and all that. Progress: added some gradients and listings based on a tutorial I saw, to make it more aesthetic. Check out the resources page!"}, {"post_id": 56998, "ts": "2026-08-29T02:05:22Z", "hours": 1.0589, "text": "added a contact page and some details now you can reach me by email and insta! (follow me pls) also I have very niche resources"}, {"post_id": 51934, "ts": "2026-08-21T21:19:00Z", "hours": 0.7114, "text": "Added Resources Slide. Said I wasn’t gonna make it all about myself, so I added a resources page (check it out it might be useful). I also added an image for my skyextendo on the project page. Hope yal enjoy it!"}, {"post_id": 50762, "ts": "2026-08-20T03:47:54Z", "hours": 0.4417, "text": "ADDED PAGES. I added an extracurricular page and a projects page. I also plan on adding study materials for sat, ap, and random stuff"}, {"post_id": 50563, "ts": "2026-08-19T19:28:47Z", "hours": 0.5008, "text": "I just made v1 of me personal site!"}]...
|
Title: My own site! by @wxie223 | Stardance
Devlo Title: My own site! by @wxie223 | Stardance
Devlogs:
I just made v1 of me personal site!
ADDED PAGES. I added an extracurricular page and a projects page. I also plan on adding study materials for sat, ap, and random stuff
Added Resources Slide. Said I wasn’t gonna make it all about myself, so I added a resources page (check it out it might be useful). I also added an image for my skyextendo on the project page. Hope yal enjoy it!
added a contact page and some details now you can reach me by email and insta! (follow me pls) also I have very niche resources
Havent been working on it a long time cuz of school and all that. Progress: added some gradients and listings based on a tutorial I saw, to make it more aesthetic. Check out the resources page!
Ships:
I shipped V1 of my personal site. Its really simple, but I hope it makes ur day.
I made a personal site about me! It has 3 pages, a main page, a projects page, and an extracurriculars page. I don’t want this website to just be all about me. I also want to include resources for highschoolers just like me, so I’m planning to add a new page with resources for like SAT, or AP, or any other useful website. Hope you like it! I made this in the personal site workshop so it is still a bit simple....
|
1793105308.0938907
|
1793969308.0938907
|
|
46784
|
My very own WebOS by @shibin | Stardance
|
shibin
|
1
|
1.54
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T14:01:57Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I built snOS, a dark glassmorphic browser personal web OS-draggable floating windows, desktop app icons, a live clock, window depth layer-ing, and a handful of functionally in-built apps: About snOS: A window with system info. Notes App Pro-a very simple notepad that saves your notes to local storage, so they don’t vanish when you refresh. Weather Forecast App: Create a live weather widget using the Open-Meteo API that allows you to search any city in the world and see live temperature, humidity, wind speeds, with dynamic condition icons. Web Browser: An inbuilt mini browser frame so you can search and browse using live web pages directly inside the desktop view. I’m a total beginner coder, so creating this without any starting point was a huge learning challenge for me. Making the window drag function feel fluid without messing up the layout was difficult. Integrating to the live weather API, processing the search box inputs, and debugging weird code errors took countless iterations. With all the times I was stuck I spent hours watching coding video tutorials, reading docs, and getting AI helpers like ChatGPT and Gemini to explain everything to me and fix my code until I finally understood how it all works. Honestly I’m just glad I didn’t give up! As a zero experience beginner seeing a blank screen turn into a functioning desktop with apps and movable window boxes was a really rewarding experience. It took patience but learning how JS, API fetching, and CSS combine to make something tangible was totally worth it. You don’t need a username or password! Open any of the desktop icons (About, Notes, Weather, or Web Browser) and use an app. Click and drag the top header of any window to move it anywhere on your screen. Click in any open window to put it on top. Other quick options: if you want to do a quick check on the current conditions for your city, try searching for your hometown in the Weather App (just hit Enter or click Search) and receive live forecast updates. Put a note in the Notes App-it auto-saves locally, so your note won’t disappear when you refresh!", "repo": "https://github.com/mshibin04/snOS", "demo": "https://sn-os-beta.vercel.app/"}]...
|
[{"post_id": 53598, "ts": "202 [{"post_id": 53598, "ts": "2026-08-24T06:52:22Z", "hours": 0.2767, "text": "upgrade!"}, {"post_id": 53587, "ts": "2026-08-24T06:13:23Z", "hours": 0.2683, "text": "My OS has one more function"}, {"post_id": 50959, "ts": "2026-08-20T13:01:07Z", "hours": 1.2217, "text": "Hey guys, I am working own my first webos"}]...
|
Title: My very own WebOS by @shibin | Stardance
D Title: My very own WebOS by @shibin | Stardance
Devlogs:
Hey guys, I am working own my first webos
My OS has one more function
upgrade!
Ships:
I built snOS, a dark glassmorphic browser personal web OS-draggable floating windows, desktop app icons, a live clock, window depth layer-ing, and a handful of functionally in-built apps: About snOS: A window with system info. Notes App Pro-a very simple notepad that saves your notes to local storage, so they don’t vanish when you refresh. Weather Forecast App: Create a live weather widget using the Open-Meteo API that allows you to search any city in the world and see live temperature, humidity, wind speeds, with dynamic condition icons. Web Browser: An inbuilt mini browser frame so you can search and browse using live web pages directly inside the desktop view. I’m a total beginner coder, so creating this without any starting point was a huge learning challenge for me. Making the window drag function feel fluid without messing up the layout was difficult. Integrating to the live weather API, processing the search box inputs, and debugging weird code errors took countless iterations. With all the times I was stuck I spent hours watching coding video tutorials, reading docs, and getting AI helpers like ChatGPT and Gemini to explain everything to me and fix my code until I finally understood how it all works. Honestly I’m just glad I didn’t give up! As a zero experience beginner seeing a blank screen turn into a functioning desktop with apps and movable window boxes was a really rewarding experience. It took patience but learning how JS, API fetching, and CSS combine to make something tangible was totally worth it. You don’t need a username or password! Open any of the desktop icons (About, Notes, Weather, or Web Browser) and use an app. Click and drag the top header of any window to move it anywhere on your screen. Click in any open window to put it on top. Other quick options: if you want to do a quick check on the current conditions for your city, try searching for your hometown in the Weather App (just hit Enter or click Search) and receive live forecast updates. Put a note in the Notes App-it auto-saves locally, so your note won’t disappear when you refresh!...
|
1793541326.353339
|
1793627726.353339
|
|
50682
|
OWLT OS by @saminsiddiqui | Stardance
|
saminsiddiqui
|
1
|
4.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T13:29:35Z", "hours": 4.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "OWLT is basically a WebOS shaped like a Deep Space Network ground station. You acquire the carrier and then three consoles come up over a starfield, and you command a probe in orbit at Jupiter. OWLT is the abbreviation for one-way light time. We type a verb into UPLINK and it is stamped with a send time, an arrival time and a resolution time in mission seconds. The rail counts the delay down, a marker crawls across the in-flight strip, and on the orbit plot you can watch the pip physically cross the gap between the two planets. Nothing on screen reports a value the spacecraft has not confirmed i.e. COMMANDED and ACKNOWLEDGED are separate readouts and they are allowed to disagree for an hour and a half. Time is compressed so that we are not actually waiting 43 minutes but compression changes how mission time maps to wall time. There is no instant mode and there never will be. The sky is generated from a seed phrase and the carrier tone is synthesised at runtime. There are four consoles so far with a settings console, an operator log, a DSKY calculator, progressive image downlink and a mission planner to come.", "repo": "https://github.com/saminsiddiqui08-beep/owlt", "demo": "https://saminsiddiqui08-beep.github.io/owlt/"}]...
|
[{"post_id": 50966, "ts": "202 [{"post_id": 50966, "ts": "2026-08-20T13:14:15Z", "hours": 0.8506, "text": "This is the final devlog for the WebOS 1 mission. The station has a voice now i.e. pressing a key at the boot gate runs a real acquisition sequence. Also, Keying the transmitter gives you a blip and forty-three minutes later the acknowledgement comes back as a two-tone chirp, so you can hear the round trip close even with every console shut. I also rebuilt the oscilloscope after realising it looked wrong for a stupid reason. The phosphor was fading about five times faster than the beam could draw so only a fifth of the trace ever existed at once. Slowing the decay to match the sweep rate fixed it instantly and it now behaves like an actual CRT with the trace dimming behind the beam instead of disappearing."}, {"post_id": 50651, "ts": "2026-08-19T22:19:34Z", "hours": 1.5347, "text": "This is the second update on OWLT. The clock is running now and the light time mechanic actually works. Type a “VERB” into UPLINK and the command takes 43 mission minutes to reach Jupiter. None of that 43 minutes is invented. Both bodies are propagated on circular orbits and the separation comes out at 777 million kilometres and dividing by the speed of light gives 43:12. It drifts down by a second every eleven thousand mission seconds because Earth is closing on Jupiter at 27 km/s."}, {"post_id": 50382, "ts": "2026-08-19T15:33:04Z", "hours": 2.0164, "text": "I basically got the shell working today. There are three draggable consoles on a seeded starfield under a fixed instrument rail and a boot gate that holds until you acquire the carrier. The part that took the longest and was the most difficult was working out the fact that a console shoved off the right edge has to be resized when the window itself resizes."}]...
|
Title: OWLT OS by @saminsiddiqui | Stardance
Devl Title: OWLT OS by @saminsiddiqui | Stardance
Devlogs:
I basically got the shell working today. There are three draggable consoles on a seeded starfield under a fixed instrument rail and a boot gate that holds until you acquire the carrier. The part that took the longest and was the most difficult was working out the fact that a console shoved off the right edge has to be resized when the window itself resizes.
This is the second update on OWLT. The clock is running now and the light time mechanic actually works. Type a “VERB” into UPLINK and the command takes 43 mission minutes to reach Jupiter. None of that 43 minutes is invented. Both bodies are propagated on circular orbits and the separation comes out at 777 million kilometres and dividing by the speed of light gives 43:12. It drifts down by a second every eleven thousand mission seconds because Earth is closing on Jupiter at 27 km/s.
This is the final devlog for the WebOS 1 mission. The station has a voice now i.e. pressing a key at the boot gate runs a real acquisition sequence. Also, Keying the transmitter gives you a blip and forty-three minutes later the acknowledgement comes back as a two-tone chirp, so you can hear the round trip close even with every console shut. I also rebuilt the oscilloscope after realising it looked wrong for a stupid reason. The phosphor was fading about five times faster than the beam could draw so only a fifth of the trace ever existed at once. Slowing the decay to match the sweep rate fixed it instantly and it now behaves like an actual CRT with the trace dimming behind the beam instead of disappearing.
Ships:
OWLT is basically a WebOS shaped like a Deep Space Network ground station. You acquire the carrier and then three consoles come up over a starfield, and you command a probe in orbit at Jupiter. OWLT is the abbreviation for one-way light time. We type a verb into UPLINK and it is stamped with a send time, an arrival time and a resolution time in mission seconds. The rail counts the delay down, a marker crawls across the in-flight strip, and on the orbit plot you can watch the pip physically cross the gap between the two planets. Nothing on screen reports a value the spacecraft has not confirmed i.e. COMMANDED and ACKNOWLEDGED are separate readouts and they are allowed to disagree for an hour and a half. Time is compressed so that we are not actually waiting 43 minutes but compression changes how mission time maps to wall time. There is no instant mode and there never will be. The sky is generated from a seed phrase and the carrier tone is synthesised at runtime. There are four consoles so far with a settings console, an operator log, a DSKY calculator, progressive image downlink and a mission planner to come....
|
1793562555.2160008
|
1793648955.2160008
|
|
49308
|
Blurt by @srihariramesh | Stardance
|
srihariramesh
|
1
|
4.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T13:24:40Z", "hours": 4.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "created Blurt a mobile which honestly made my own life a lot easier it is basically an app where the user speaks into it and the app auto-transcribes and gets the intent out of the user’s speech. It says what it saves as reminders, saves things that the user needs to remember, and cuts off the blurting part of the user. It says it saves the important part that the user has to remember. The most challenging part was about UI. I didn’t have any idea specifically but somehow found out some kind of good UI. I’m looking forward to it being like ChatGPT voice level. There are only two devlogs because I had started this a lot earlier, at the start of 2026, to actually build this but when I got to know about stardance, it was this late. Thanks guys and I really hope my app does well for you too. If you try it out, don’t forget to add your keys in the settings or else it doesn’t work.", "repo": "https://github.com/srihariramesh2023/blurt", "demo": "https://blurt-shr.netlify.app/"}]...
|
[{"post_id": 50967, "ts": "202 [{"post_id": 50967, "ts": "2026-08-20T13:17:21Z", "hours": 3.1828, "text": "it want how i expected it to be soo gotta work and im still stuck at the same screen but changing the backend but if anyone has an ideas please do share…"}, {"post_id": 48090, "ts": "2026-08-16T11:42:23Z", "hours": 0.7264, "text": "trying on improving the library tab on BLURT"}]...
|
Title: Blurt by @srihariramesh | Stardance
Devlog Title: Blurt by @srihariramesh | Stardance
Devlogs:
trying on improving the library tab on BLURT
it want how i expected it to be soo gotta work and im still stuck at the same screen but changing the backend but if anyone has an ideas please do share…
Ships:
created Blurt a mobile which honestly made my own life a lot easier it is basically an app where the user speaks into it and the app auto-transcribes and gets the intent out of the user’s speech. It says what it saves as reminders, saves things that the user needs to remember, and cuts off the blurting part of the user. It says it saves the important part that the user has to remember. The most challenging part was about UI. I didn’t have any idea specifically but somehow found out some kind of good UI. I’m looking forward to it being like ChatGPT voice level. There are only two devlogs because I had started this a lot earlier, at the start of 2026, to actually build this but when I got to know about stardance, it was this late. Thanks guys and I really hope my app does well for you too. If you try it out, don’t forget to add your keys in the settings or else it doesn’t work....
|
1793541389.9049923
|
1793627789.9049923
|
|
50857
|
Aur helper & Installer by @Aashmn13 | Stardanc Aur helper & Installer by @Aashmn13 | Stardance...
|
Aashmn13
|
1
|
3.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T12:48:02Z", "hours": 3.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "It is interactive installer for arch which installs it from AUR. I made some changes like better search, cache management and removes many packages at once. This project was one of the difficult scripts i ever made, it includes various unknown functions which was unknown to me before. The most difficult part was to make the interative menu.", "repo": "https://github.com/Aashman16wi/my-scripts/blob/main/AUR-Helper%20Search%20%26%20Management%20Utility", "demo": ""}]...
|
[{"post_id": 50933, "ts": "202 [{"post_id": 50933, "ts": "2026-08-20T12:25:19Z", "hours": 2.5681, "text": "I have Successfully updatedd my script having better search, Cache management, Multiple package remover and fixed the colours. This script is only usable for Arch or Arch-based like Endeavour OS and Cachy OS."}]...
|
Title: Aur helper & Installer by @Aashmn13 | S Title: Aur helper & Installer by @Aashmn13 | Stardance
Devlogs:
I have Successfully updatedd my script having better search, Cache management, Multiple package remover and fixed the colours. This script is only usable for Arch or Arch-based like Endeavour OS and Cachy OS.
Ships:
It is interactive installer for arch which installs it from AUR. I made some changes like better search, cache management and removes many packages at once. This project was one of the difficult scripts i ever made, it includes various unknown functions which was unknown to me before. The most difficult part was to make the interative menu....
|
1793407714.3454852
|
1793494114.3454852
|
|
50622
|
Quary by @Veert | Stardance
|
Veert
|
1
|
141.33
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T12:28:09Z", "hours": 14.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Working on improving the bot behavior aswell as mastering the gravity effects", "repo": "https://github.com/VrSchool11/Quary", "demo": "https://vrschool11.github.io/Quary/"}]...
|
[{"post_id": 67746, "ts": "202 [{"post_id": 67746, "ts": "2026-09-13T23:58:49Z", "hours": 5.4667, "text": "Rewrote Text and captions to sound more human"}, {"post_id": 66299, "ts": "2026-09-12T01:54:14Z", "hours": 13.8675, "text": "Making the game less ai by rewriting the affected parts"}, {"post_id": 65056, "ts": "2026-09-09T21:55:44Z", "hours": 8.4336, "text": "Adding some new features and improving on gameplay and improving the performance of the game on low-end devices"}, {"post_id": 64787, "ts": "2026-09-09T13:29:53Z", "hours": 4.6419, "text": "Working one some gameplay fixed and fixing some bugs"}, {"post_id": 64540, "ts": "2026-09-09T01:07:14Z", "hours": 1.0333, "text": "Completely Rewrote my code as when i shipped it the reviewer said it was over 30% ai now fixed it and its now below 30% ai"}, {"post_id": 64193, "ts": "2026-09-08T12:50:07Z", "hours": 7.6981, "text": "Fixing General Bugs and Fixing The boss fight"}, {"post_id": 63477, "ts": "2026-09-07T11:09:45Z", "hours": 8.8664, "text": "Working On Fixing The readability of the captions and fixing the tips and the tutorial for the game along with improving the layout of the game"}, {"post_id": 62152, "ts": "2026-09-05T12:26:37Z", "hours": 9.0681, "text": "Guns are in. You take one off the first guard you kill, and firing it wakes the whole zone so you build a silencer out of a sleeve, a spring and an oily rag. The king’s an actual fight now too, he winds up before every move so you can dodge it and one of them rips a limb off you and chucks it on the floor. He used to one-shot you if you’d gone full animal, so that’s gone. Checkpoints on every zone now, plus a pile of daft bugs fixed."}, {"post_id": 61860, "ts": "2026-09-04T23:59:57Z", "hours": 4.2339, "text": "Binned the black hole game after about a day and started Quarry instead. You’re a scientist who got kidnapped for something you built, they took your limbs, and you escape by tearing limbs off whatever you can find. Animal ones jump higher and hit harder, but the guards can see you’re not human anymore, which is basically the whole game. First playable version was rubbish. Every gap was too wide to clear, so you couldn’t actually finish it. Since then I’ve added five zones, a king to fight at the end, guns you loot off dead guards, and a silencer you build out of a sleeve, a spring and an oiled rag. Also spent way too long moving the gun sprite up four pixels because at hip height it looked like something else."}, {"post_id": 61283, "ts": "2026-09-04T04:06:54Z", "hours": 2.9639, "text": "Changed the Game Completely and moved onto another game idea as the other one didn’t seem to be working out heres a sneak peek below"}, {"post_id": 59498, "ts": "2026-09-01T11:27:27Z", "hours": 6.7897, "text": "1 September Today i completely redesigned the UI and also added two new features Different Type of foods A SpeedRun mode On top of that cleaned up the repo and fixed some minor issues here and there"}, {"post_id": 57370, "ts": "2026-08-29T14:27:55Z", "hours": 12.195, "text": "aug 29 completely remade the banner today — scrapped the old one and rebuilt it from scratch so it actually holds together. after that i cleaned up the code, tidying things up and cutting out the rough bits, and did some general improvements across the game while i was in there. more features coming soon."}, {"post_id": 54428, "ts": "2026-08-25T13:07:06Z", "hours": 9.8839, "text": "Fixed the big bug where growing your hole past a certain size would whiteout the entire screen — turned out the camera and the glow effects were both scaling wrong at high mass. While I was in there I added a bunch of new stuff: kill cam replays when you die, comets and binary planets that mess with gravity, a tracker that shows when a rogue hole is hunting you, and permanent achievements on top of the regular quests. Hosts can also now tweak gravity strength, arena size, and bot count before opening a station. Caught and fixed a couple of sneaky bugs along the way too, like eat sounds losing their punch on big absorbs. It’s all live on the GitHub Pages link now."}, {"post_id": 52329, "ts": "2026-08-22T13:13:51Z", "hours": 7.4672, "text": "Turned out if you’d already played a round and went back to the menu, you’d hear every bot in the background eating dust at full volume — the sound code treated “nobody’s watching” as “max volume” instead of silent. Fixed.Also added a pile of new stuff to actually do out there: dash (E key, quick burst forward, costs mass), wormholes that teleport you across the map, random supernova blasts that shove everything outward, a gravity slingshot off planets for a free speed boost, slow mass decay if you’re running away with the lead, a shockwave when you eat something way smaller than you, and a token bounty on whoever’s currently #1."}, {"post_id": 52060, "ts": "2026-08-22T03:27:35Z", "hours": 14.3667, "text": "Spent this session on polish and cleanup. Reworked the UI and CSS myself — set up a color palette with variables so the whole game re-themes from one place, and simplified a lot of the styling. Also fixed a performance bug: the game ran fine in Safari but lagged in Chrome. Turned out canvas shadow-blur is way slower in Chromium, so I cut it off the small stuff (dust, particles) and leaned on the additive glow instead. Runs smooth now. Still fixing the plinko bonus round and cleaning things up. More soon."}, {"post_id": 51642, "ts": "2026-08-21T12:55:44Z", "hours": 4.9736, "text": "Big redesign + feature passRepainted the whole UI — it was reading way too “safe dark-mode template,” so it’s now themed around eclipse observation expeditions instead: brass, khaki, oxide colors, stencil title font, manifest-style panels. Gameplay’s untouched, just the look changed.Also shipped a bunch of stuff that had been on the list:personal best (mass + survival time) on the menusettings: quality toggle, volume slider, minimap on/offkill streak popup when you’re on a rollfree color picker for your holefirst-time tutorial popupauto-lowers quality if your fps tankspower-ups: speed, magnet, shieldtimed matches and last-hole-standing modeteams (red vs blue), no friendly firedaily challenge — same seeded world for everyone each dayFound and fixed two real bugs while testing: the shop wasn’t updating your token count after a purchase, and timed matches were miscalculating rank when multiple people hit the clock at once."}, {"post_id": 51436, "ts": "2026-08-21T04:27:59Z", "hours": 3.7672, "text": "UPDATE — CSS rebuild + progression system. Review sent it back saying my CSS looked too AI-made, so I rebuilt the styling myself and actually learned how it works — set up a color palette with variables and re-themed the whole game in one line to prove it. While I was in there I added a progression system: tokens you earn each run and keep between games, side quests for bonus tokens, a cosmetics shop (colors and orbiting shapes, no pay-to-win), and an end-of-run plinko round that multiplies your run tokens (minimum 1x, so you can never lose them), plus arrow-key/WASD movement and a pause button. Biggest headache was the plinko ball not landing in a slot — the physics were using the display size instead of the canvas’s real size — fixed once I matched them."}, {"post_id": 50987, "ts": "2026-08-20T14:02:44Z", "hours": 1.6131, "text": "UPDATE — Multiplayer is live (online, with room codes) Summary Blackhole.io is now real online multiplayer. Create a room, share the code, and someone on another device joins and you can actually eat each other. Not bots — real people. Breakdown moved the whole game simulation onto a server so it’s authoritative (server decides what’s real, clients can’t cheat) clients now only send their mouse target; the server sends back everyone’s positions create/join a room with a short code, each room is its own isolated game added interpolation so other players move smoothly instead of teleporting between updates bots still fill empty rooms so it’s never a dead lobby deployed the server to Render and the game page to GitHub Pages, wired together over WebSockets What fought me The networking logic wasn’t even the hard part — deploying it was. Getting the GitHub Pages page to actually talk to the Render server took forever. The site kept opening my README instead of the game, the WebSocket wouldn’t connect because it was still pointing at localhost, and it had to be wss:// not ws:// or the browser silently blocked it. Bunch of small things that each looked like the whole thing was broken. Once the client finally connected to the live server and I saw a second black hole move on my screen, that was the moment. Note: Server May Take some time to start up"}, {"post_id": 50934, "ts": "2026-08-20T12:25:40Z", "hours": 6.6444, "text": "Working On fixing the lag caused by the advance gravity effects"}, {"post_id": 50512, "ts": "2026-08-19T18:02:36Z", "hours": 7.405, "text": "Spent today getting Blackhole.io off the ground. You’re a black hole — you eat dust, grow, and start pulling in bigger stuff. Got the core loop working: move toward the mouse, swallow anything smaller, get bigger. Made growth area-based so it doesn’t snowball out of control in ten seconds. The part that ate my time was gravity. I didn’t want stuff to just slide toward me — I wanted it to actually orbit and spiral in. Took a lot of fiddling with the numbers before it felt right instead of floaty. About 7.5 hours in so far. Next up: making the other black holes smarter so it’s not just me alone out there aswell as a multiplayer version so you can play with your friends"}]...
|
Title: Quary by @Veert | Stardance
Devlogs:
Spent Title: Quary by @Veert | Stardance
Devlogs:
Spent today getting Blackhole.io off the ground. You’re a black hole — you eat dust, grow, and start pulling in bigger stuff. Got the core loop working: move toward the mouse, swallow anything smaller, get bigger. Made growth area-based so it doesn’t snowball out of control in ten seconds. The part that ate my time was gravity. I didn’t want stuff to just slide toward me — I wanted it to actually orbit and spiral in. Took a lot of fiddling with the numbers before it felt right instead of floaty. About 7.5 hours in so far. Next up: making the other black holes smarter so it’s not just me alone out there aswell as a multiplayer version so you can play with your friends
Working On fixing the lag caused by the advance gravity effects
UPDATE — Multiplayer is live (online, with room codes) Summary Blackhole.io is now real online multiplayer. Create a room, share the code, and someone on another device joins and you can actually eat each other. Not bots — real people. Breakdown moved the whole game simulation onto a server so it’s authoritative (server decides what’s real, clients can’t cheat) clients now only send their mouse target; the server sends back everyone’s positions create/join a room with a short code, each room is its own isolated game added interpolation so other players move smoothly instead of teleporting between updates bots still fill empty rooms so it’s never a dead lobby deployed the server to Render and the game page to GitHub Pages, wired together over WebSockets What fought me The networking logic wasn’t even the hard part — deploying it was. Getting the GitHub Pages page to actually talk to the Render server took forever. The site kept opening my README instead of the game, the WebSocket wouldn’t connect because it was still pointing at localhost, and it had to be wss:// not ws:// or the browser silently blocked it. Bunch of small things that each looked like the whole thing was broken. Once the client finally connected to the live server and I saw a second black hole move on my screen, that was the moment. Note: Server May Take some time to start up
UPDATE — CSS rebuild + progression system. Review sent it back saying my CSS looked too AI-made, so I rebuilt the styling myself and actually learned how it works — set up a color palette with variables and re-themed the whole game in one line to prove it. While I was in there I added a progression system: tokens you earn each run and keep between games, side quests for bonus tokens, a cosmetics shop (colors and orbiting shapes, no pay-to-win), and an end-of-run plinko round that multiplies your run tokens (minimum 1x, so you can never lose them), plus arrow-key/WASD movement and a pause button. Biggest headache was the plinko ball not landing in a slot — the physics were using the display size instead of the canvas’s real size — fixed once I matched them.
Big redesign + feature passRepainted the whole UI — it was reading way too “safe dark-mode template,” so it’s now themed around eclipse observation expeditions instead: brass, khaki, oxide colors, stencil title font, manifest-style panels. Gameplay’s untouched, just the look changed.Also shipped a bunch of stuff that had been on the list:personal best (mass + survival time) on the menusettings: quality toggle, volume slider, minimap on/offkill streak popup when you’re on a rollfree color picker for your holefirst-time tutorial popupauto-lowers quality if your fps tankspower-ups: speed, magnet, shieldtimed matches and last-hole-standing modeteams (red vs blue), no friendly firedaily challenge — same seeded world for everyone each dayFound and fixed two real bugs while testing: the shop wasn’t updating your token count after a purchase, and timed matches were miscalculating rank when multiple people hit the clock at once.
Spent this session on polish and cleanup. Reworked the UI and CSS myself — set up a color palette with variables so the whole game re-themes from one place, and simplified a lot of the styling. Also fixed a performance bug: the game ran fine in Safari but lagged in Chrome. Turned out canvas shadow-blur is way slower in Chromium, so I cut it off the small stuff (dust, particles) and leaned on the additive glow instead. Runs smooth now. Still fixing the plinko bonus round and cleaning things up. More soon.
Turned out if you’d already played a round and went back to the menu, you’d hear every bot in the background eating dust at full volume — the sound code treated “nobody’s watching” as “max volume” instead of silent. Fixed.Also added a pile of new stuff to actually do out there: dash (E key, quick burst forward, costs mass), wormholes that teleport you across the map, random supernova blasts that shove everything outward, a gravity slingshot off planets for a free speed boost, slow mass decay if you’re running away with the lead, a shockwave when you eat something way smaller than you, and a token bounty on whoever’s currently #1.
Fixed the big bug where growing your hole past a certain size would whiteout the entire screen — turned out the camera and the glow effects were both scaling wrong at high mass. While I was in there I added a bunch of new stuff: kill cam replays when you die, comets and binary planets that mess with gravity, a tracker that shows when a rogue hole is hunting you, and permanent achievements on top of the regular quests. Hosts can also now tweak gravity strength, arena size, and bot count before opening a station. Caught and fixed a couple of sneaky bugs along the way too, like eat sounds losing their punch on big absorbs. It’s all live on the GitHub Pages link now.
aug 29 completely remade the banner today — scrapped the old one and rebuilt it from scratch so it actually holds together. after that i cleaned up the code, tidying things up and cutting out the rough bits, and did some general improvements across the game while i was in there. more features coming soon.
1 September Today i completely redesigned the UI and also added two new features Different Type of foods A SpeedRun mode On top of that cleaned up the repo and fixed some minor issues here and there
Changed the Game Completely and moved onto another game idea as the other one didn’t seem to be working out heres a sneak peek below
Binned the black hole game after about a day and started Quarry instead. You’re a scientist who got kidnapped for something you built, they took your limbs, and you escape by tearing limbs off whatever you can find. Animal ones jump higher and hit harder, but the guards can see you’re not human anymore, which is basically the whole game. First playable version was rubbish. Every gap was too wide to clear, so you couldn’t actually finish it. Since then I’ve added five zones, a king to fight at the end, guns you loot off dead guards, and a silencer you build out of a sleeve, a spring and an oiled rag. Also spent way too long moving the gun sprite up four pixels because at hip height it looked like something else.
Guns are in. You take one off the first guard you kill, and firing it wakes the whole zone so you build a silencer out of a sleeve, a spring and an oily rag. The king’s an actual fight now too, he winds up before every move so you can dodge it and one of them rips a limb off you and chucks it on the floor. He used to one-shot you if you’d gone full animal, so that’s gone. Checkpoints on every zone now, plus a pile of daft bugs fixed.
Working On Fixing The readability of the captions and fixing the tips and the tutorial for the game along with improving the layout of the game
Fixing General Bugs and Fixing The boss fight
Completely Rewrote my code as when i shipped it the reviewer said it was over 30% ai now fixed it and its now below 30% ai
Working one some gameplay fixed and fixing some bugs
Adding some new features and improving on gameplay and improving the performance of the game on low-end devices
Making the game less ai by rewriting the affected parts
Rewrote Text and captions to sound more human
Ships:
Working on improving the bot behavior aswell as mastering the gravity effects...
|
1793545004.7063923
|
1793631404.7063923
|
|
51646
|
My own site! by @lukas_56376 | Stardance
|
lukas_56376
|
1
|
2.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T12:11:29Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Hello, this is my first project here and also my first website, without major Ai usage, the only AI usage was debugging the CSS, everything else island written!", "repo": "https://github.com/Lukas-56376/My-Website---Lukas", "demo": "https://lukas-56376.github.io/My-Website---Lukas/"}]...
|
[{"post_id": 50906, "ts": "202 [{"post_id": 50906, "ts": "2026-08-20T11:41:22Z", "hours": 1.5767, "text": "Heya, just finished V1 of my Website, do you have any ides or suggestions?"}]...
|
Title: My own site! by @lukas_56376 | Stardance
D Title: My own site! by @lukas_56376 | Stardance
Devlogs:
Heya, just finished V1 of my Website, do you have any ides or suggestions?
Ships:
Hello, this is my first project here and also my first website, without major Ai usage, the only AI usage was debugging the CSS, everything else island written!...
|
1792878556.5340533
|
1793742556.5340533
|
|
51344
|
Aarav Workbench OS by @Aaravkatiyar | Stardance
|
Aaravkatiyar
|
1
|
2.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T11:59:45Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Aarav Workbench OS is a public browser workbench built around a calmer project desk. Visitors can open, drag, focus, minimize, restore, and maximize separate windows; use the Command Dock; write browser-local notes; choose themes; save a workspace snapshot; and try two offline games. The latest update adds Clear Desk: it returns the desktop to the landing screen without deleting the visitor’s local notes, journal text, saved layout, themes, focus-list steps, or scores. I also corrected the Console switching edge case so it does not claim a window switch when no app is open. I built this on a limited laptop setup while preparing for exams in about two weeks, so I focused on useful browser-first features and tested each release carefully. JavaScript checks, local Console and Command Dock flows, and a fresh GitHub Pages check all passed without console errors. Try it: https://aaravkatiyar55-gif.github.io/aarav-workbench-os/ Source: https://github.com/aaravkatiyar55-gif/aarav-workbench-os/", "repo": "https://github.com/aaravkatiyar55-gif/aarav-workbench-os", "demo": "https://aaravkatiyar55-gif.github.io/aarav-workbench-os/"}]...
|
[{"post_id": 50911, "ts": "202 [{"post_id": 50911, "ts": "2026-08-20T11:44:09Z", "hours": 0.3314, "text": "I added a small desktop-cleanup feature to Aarav Workbench OS: Clear Desk.It is for the moment when a visitor has opened a few tools and wants to get back to the calm landing screen. It hides the visible windows, but it does not delete browser-local notes, journal text, themes, focus-list steps, scores, or saved workspace data. The dock can restore any app afterwards.I also tightened the window-switching edge case so Console next / previous does not claim a switch when no window is open. I ran the JavaScript checks, tested Clear Desk through the browser-only Console and Command Dock locally, then opened the fresh GitHub Pages release again. The deployed action returned to landing with zero visible windows and no browser-console errors.Live demo: https://aaravkatiyar55-gif.github.io/aarav-workbench-os/Source : https://github.com/aaravkatiyar55-gif/aarav-workbench-osIf you try it, what would make the desk easier to return to: one shortcut, a clearer dock state, or a different landing action?"}, {"post_id": 50883, "ts": "2026-08-20T10:43:33Z", "hours": 0.4025, "text": "Today I worked on the next Aarav Workbench OS iteration and focused on making the desktop easier to use instead of just adding more windows.The desk starts clean, then visitors can open what they need from the dock or choose an Explore, Focus, or Break desk mode. I added local-only tools including a small Notebook, Focus List, Theme Lab, Workspace Snapshot, and a Game Room with two offline games: Signal Sprint and Desk Grid.The latest UI pass also adds a desktop window strip so an open or minimized app can be focused or restored without going back to the landing screen. I tested the open → minimize → restore flow locally, ran the JavaScript checks, and verified the latest GitHub Pages build loads without console errors.Everything saved by these tools stays in the visitor’s browser; there is no account or cloud sync. What would make this workbench more useful next: better project browsing, another practical local tool, or improvements to the games?"}, {"post_id": 50579, "ts": "2026-08-19T19:40:25Z", "hours": 1.3503, "text": "I built the desktop foundation for Aarav Workbench OS: a project-desk interface with a top bar, dock, draggable and maximizable windows, plus keyboard-friendly window controls. The app opens to a clean landing view so visitors can choose what to explore instead of arriving at a cluttered desktop.I also added a Command Dock and saved workspace snapshot so the desktop can be focused, arranged, and restored after refresh. The workbench is intentionally a different concept and visual system from the guide.What should I improve next: a better notes workflow, more theme choices, or a new utility?"}]...
|
Title: Aarav Workbench OS by @Aaravkatiyar | Stard Title: Aarav Workbench OS by @Aaravkatiyar | Stardance
Devlogs:
I built the desktop foundation for Aarav Workbench OS: a project-desk interface with a top bar, dock, draggable and maximizable windows, plus keyboard-friendly window controls. The app opens to a clean landing view so visitors can choose what to explore instead of arriving at a cluttered desktop.I also added a Command Dock and saved workspace snapshot so the desktop can be focused, arranged, and restored after refresh. The workbench is intentionally a different concept and visual system from the guide.What should I improve next: a better notes workflow, more theme choices, or a new utility?
Today I worked on the next Aarav Workbench OS iteration and focused on making the desktop easier to use instead of just adding more windows.The desk starts clean, then visitors can open what they need from the dock or choose an Explore, Focus, or Break desk mode. I added local-only tools including a small Notebook, Focus List, Theme Lab, Workspace Snapshot, and a Game Room with two offline games: Signal Sprint and Desk Grid.The latest UI pass also adds a desktop window strip so an open or minimized app can be focused or restored without going back to the landing screen. I tested the open → minimize → restore flow locally, ran the JavaScript checks, and verified the latest GitHub Pages build loads without console errors.Everything saved by these tools stays in the visitor’s browser; there is no account or cloud sync. What would make this workbench more useful next: better project browsing, another practical local tool, or improvements to the games?
I added a small desktop-cleanup feature to Aarav Workbench OS: Clear Desk.It is for the moment when a visitor has opened a few tools and wants to get back to the calm landing screen. It hides the visible windows, but it does not delete browser-local notes, journal text, themes, focus-list steps, scores, or saved workspace data. The dock can restore any app afterwards.I also tightened the window-switching edge case so Console next / previous does not claim a switch when no window is open. I ran the JavaScript checks, tested Clear Desk through the browser-only Console and Command Dock locally, then opened the fresh GitHub Pages release again. The deployed action returned to landing with zero visible windows and no browser-console errors.Live demo: https://aaravkatiyar55-gif.github.io/aarav-workbench-os/Source : https://github.com/aaravkatiyar55-gif/aarav-workbench-osIf you try it, what would make the desk easier to return to: one shortcut, a clearer dock state, or a different landing action?
Ships:
Aarav Workbench OS is a public browser workbench built around a calmer project desk. Visitors can open, drag, focus, minimize, restore, and maximize separate windows; use the Command Dock; write browser-local notes; choose themes; save a workspace snapshot; and try two offline games. The latest update adds Clear Desk: it returns the desktop to the landing screen without deleting the visitor’s local notes, journal text, saved layout, themes, focus-list steps, or scores. I also corrected the Console switching edge case so it does not claim a window switch when no app is open. I built this on a limited laptop setup while preparing for exams in about two weeks, so I focused on useful browser-first features and tested each release carefully. JavaScript checks, local Console and Command Dock flows, and a fresh GitHub Pages check all passed without console errors. Try it: https://aaravkatiyar55-gif.github.io/aarav-workbench-os/ Source: https://github.com/aaravkatiyar55-gif/aarav-workbench-os/...
|
1793486940.2868843
|
1793573340.2868843
|
|
42247
|
Compose ADB GUI by @door | Stardance
|
door
|
1
|
23.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T11:38:24Z", "hours": 23.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I built an app that lets you easily manage your Android device! It includes a package manager, file manager, terminal, and quick actions that allow you to get uptime, screen resolution, reboot into bootloader, reboot… You can also Remove all of the annoying bloatware Samsung or whatever manuacturer you have by simply selecting your device, selecting which manufacturer the device is from and then clicking Run!", "repo": "https://github.com/whatAdoor/ADB-GUI", "demo": "https://github.com/whatAdoor/ADB-GUI/releases"}]...
|
[{"post_id": 50894, "ts": "202 [{"post_id": 50894, "ts": "2026-08-20T11:07:23Z", "hours": 1.7222, "text": "Just finished. I set up github actions to build the app and generated a app icon with Gemini. I had a hard time setting it tho because i edited the wrong gradle file."}, {"post_id": 50526, "ts": "2026-08-19T18:29:15Z", "hours": 3.62, "text": "Implemented the package manager, you can search for your apps and delete them, system apps included. You can also click the little filter Icon on the top right corner and select a manufacturer which will then select the appropriate tasks to remove bloatware and preinstalled apps that nobody needs."}, {"post_id": 47637, "ts": "2026-08-15T17:45:50Z", "hours": 6.5169, "text": "Finished the File Manager, but i had to strip down its functionality a bit because im still learning and I dont want to just vibe code it. So now you can browse the files, delete, download and upload files in the File Manager. Next I will make the Terminal and Package Manager and the different actions and I think that will wrap up this project then."}, {"post_id": 41490, "ts": "2026-08-05T18:48:04Z", "hours": 4.0931, "text": "First steps on the File Manager done, I am basically just using ls and then interpreting each line and turning them into data objects to then display them. Gonna start to implement the Actions like download, upload, copy… tomorrow. I also feel like changing up the Icons a bit, and maybe I’ll start switching up the colors a bit and experimenting with that. 👀"}, {"post_id": 40317, "ts": "2026-08-03T19:12:17Z", "hours": 0.6756, "text": "Day 3: Today i was short on time so i just added two Buttons, which will allow the user to open a terminal to a specific device, and one that will open a file Explorer to a specific device. I also polished the output Box formatting. The next big task will be to implement the File Explorer, after that I will add more commands to the command list. I also thought about adding a third button where you can see all the installed packages on a device and uninstall them. Currently you can select devices connected via adb and then select commands you want to be executed on the selected device(s) and click run, the output will be shown in the black box."}, {"post_id": 39557, "ts": "2026-08-02T16:21:37Z", "hours": 2.7569, "text": "Day 2: Did some polishing on the UI and implemented a search bar. Please leave reccomendations for the commands i should implement in the comments!"}, {"post_id": 39103, "ts": "2026-08-01T20:13:48Z", "hours": 3.2736, "text": "Brainstorming on the core functionality and implemented some first layouts. Doesnt look good but rebooting works already."}]...
|
Title: Compose ADB GUI by @door | Stardance
Devlo Title: Compose ADB GUI by @door | Stardance
Devlogs:
Brainstorming on the core functionality and implemented some first layouts. Doesnt look good but rebooting works already.
Day 2: Did some polishing on the UI and implemented a search bar. Please leave reccomendations for the commands i should implement in the comments!
Day 3: Today i was short on time so i just added two Buttons, which will allow the user to open a terminal to a specific device, and one that will open a file Explorer to a specific device. I also polished the output Box formatting. The next big task will be to implement the File Explorer, after that I will add more commands to the command list. I also thought about adding a third button where you can see all the installed packages on a device and uninstall them. Currently you can select devices connected via adb and then select commands you want to be executed on the selected device(s) and click run, the output will be shown in the black box.
First steps on the File Manager done, I am basically just using ls and then interpreting each line and turning them into data objects to then display them. Gonna start to implement the Actions like download, upload, copy… tomorrow. I also feel like changing up the Icons a bit, and maybe I’ll start switching up the colors a bit and experimenting with that. 👀
Finished the File Manager, but i had to strip down its functionality a bit because im still learning and I dont want to just vibe code it. So now you can browse the files, delete, download and upload files in the File Manager. Next I will make the Terminal and Package Manager and the different actions and I think that will wrap up this project then.
Implemented the package manager, you can search for your apps and delete them, system apps included. You can also click the little filter Icon on the top right corner and select a manufacturer which will then select the appropriate tasks to remove bloatware and preinstalled apps that nobody needs.
Just finished. I set up github actions to build the app and generated a app icon with Gemini. I had a hard time setting it tho because i edited the wrong gradle file.
Ships:
I built an app that lets you easily manage your Android device! It includes a package manager, file manager, terminal, and quick actions that allow you to get uptime, screen resolution, reboot into bootloader, reboot… You can also Remove all of the annoying bloatware Samsung or whatever manuacturer you have by simply selecting your device, selecting which manufacturer the device is from and then clicking Run!...
|
1793432903.4481688
|
1793519303.4481688
|
|
40525
|
Autonomous Rescue & Extraction Drone 3D Design Autonomous Rescue & Extraction Drone 3D Design by @jeremyngai | Stardance...
|
jeremyngai
|
1
|
16.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T11:28:37Z", "hours": 16.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I modelled this 8 rotor drone in solid edge. it is designed to safely guide people out of disaster like flooding or earthquake. How it works? The drone has 3 main sensors, lidar (terrain mapping), uwb radar (human heartbeat), and thermal imaging (body heat), the ai fuses these sensors to locate victim even in absolute terrible weather like storms or smoke. Once the drone has found victim, it lowers a full body harness with a rope to grab the victim and safely move them to safe area. the body harness is equipped with 6.3 inch display, which allow the first responder to provide immediate assistant for the victim using lora communication. The body harness also has built in heart rate sensor and blood saturation sensor in the body mesh, so that the first responder can immediately diagnose the victim. To make repairs and maintainence convenient, there is a hot swappable battery drawer and flight controller drawer on the sides so maintainer can make instant repair for the next round. This is my first time modelling complex 3d design like this and it turned out way better than i expected lol. so there is definitely alot of things i just learnt and its just fun to design this drone. i thought model like this is impossible for me to make at first. since i cant work on this at school, i had to grind this at home and during weekends lol. To have a look at this model, i have converted it into a .obj file and uploaded into sketchfab so have a look!! (its in the ‘video of project’ button). However if you want to see every bits of the project you can download all the files on my github repo and install solid edge to open the assembly file. i have also made a pdf slides explaning the model in super detail in my github repo, SO HAVE A LOOK!!", "repo": "https://github.com/ngai-jeremy/Autonomous-Rescue-Extraction-Drone", "demo": ""}]...
|
[{"post_id": 50214, "ts": "202 [{"post_id": 50214, "ts": "2026-08-19T10:58:00Z", "hours": 2.5503, "text": "I completed my 3d design slides presentation!! this is the last step of this project. I’ll be shipping soon!!"}, {"post_id": 47384, "ts": "2026-08-15T09:17:05Z", "hours": 4.2833, "text": "Over the past few days, i worked on the final design of the drone! These are the things added/changed I finalised the propeller design with aero curve for all 8 propellers. I added colour to all the parts. I added a landing gear skid on the bottom for landing I finalised the assembly file I’m now working on the presentation slides. so stay tuned!! its so fun"}, {"post_id": 43731, "ts": "2026-08-09T04:24:43Z", "hours": 3.3272, "text": "Over the past few days i worked on the body harness with a D ring and made the battery compartment. the propeller is dummy right now for size inspection"}, {"post_id": 40697, "ts": "2026-08-04T12:01:25Z", "hours": 0.9667, "text": "Added rope, full body seat belt soon!"}, {"post_id": 39350, "ts": "2026-08-02T08:40:03Z", "hours": 4.0358, "text": "Spent the last 2 days modelling the main chassis design and the 42 inch propeller and its a fun journey lol. I’ve never designed something this complex. That hole is for the battery which can be replaced out. Then on the front face there’s sensor and cameras for UWB Radar for life detection and LiDAR camera for terrain mapping, and also lights."}, {"post_id": 38643, "ts": "2026-08-01T04:31:28Z", "hours": 0.6892, "text": "Made a rough draft of the drone design in Solid Edge software"}]...
|
Title: Autonomous Rescue & Extraction Drone 3D Title: Autonomous Rescue & Extraction Drone 3D Design by @jeremyngai | Stardance
Devlogs:
Made a rough draft of the drone design in Solid Edge software
Spent the last 2 days modelling the main chassis design and the 42 inch propeller and its a fun journey lol. I’ve never designed something this complex. That hole is for the battery which can be replaced out. Then on the front face there’s sensor and cameras for UWB Radar for life detection and LiDAR camera for terrain mapping, and also lights.
Added rope, full body seat belt soon!
Over the past few days i worked on the body harness with a D ring and made the battery compartment. the propeller is dummy right now for size inspection
Over the past few days, i worked on the final design of the drone! These are the things added/changed I finalised the propeller design with aero curve for all 8 propellers. I added colour to all the parts. I added a landing gear skid on the bottom for landing I finalised the assembly file I’m now working on the presentation slides. so stay tuned!! its so fun
I completed my 3d design slides presentation!! this is the last step of this project. I’ll be shipping soon!!
Ships:
I modelled this 8 rotor drone in solid edge. it is designed to safely guide people out of disaster like flooding or earthquake. How it works? The drone has 3 main sensors, lidar (terrain mapping), uwb radar (human heartbeat), and thermal imaging (body heat), the ai fuses these sensors to locate victim even in absolute terrible weather like storms or smoke. Once the drone has found victim, it lowers a full body harness with a rope to grab the victim and safely move them to safe area. the body harness is equipped with 6.3 inch display, which allow the first responder to provide immediate assistant for the victim using lora communication. The body harness also has built in heart rate sensor and blood saturation sensor in the body mesh, so that the first responder can immediately diagnose the victim. To make repairs and maintainence convenient, there is a hot swappable battery drawer and flight controller drawer on the sides so maintainer can make instant repair for the next round. This is my first time modelling complex 3d design like this and it turned out way better than i expected lol. so there is definitely alot of things i just learnt and its just fun to design this drone. i thought model like this is impossible for me to make at first. since i cant work on this at school, i had to grind this at home and during weekends lol. To have a look at this model, i have converted it into a .obj file and uploaded into sketchfab so have a look!! (its in the ‘video of project’ button). However if you want to see every bits of the project you can download all the files on my github repo and install solid edge to open the assembly file. i have also made a pdf slides explaning the model in super detail in my github repo, SO HAVE A LOOK!!...
|
1793479845.2458625
|
1793566245.2458625
|
|
51820
|
Legal Saathi by @Aaravkatiyar | Stardance
|
Aaravkatiyar
|
1
|
17.22
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T11:11:05Z", "hours": 15.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Legal Saathi is an AI-assisted legal-information and preparation workspace for English, Hindi, and Hinglish. It is designed to help people organise facts, understand safer next steps, review documents, and prepare draft structures — not to file, send, pay, contact anyone, or replace an advocate. I began iterating on this project before I knew about Stardance and joined the challenge recently, so the 15 hours shown here are the time currently recorded by the linked tracker. They are separate from the longer build journey. I have been building on limited hardware, which made testing and iteration slower but pushed me to focus on clearer flows, safer boundaries, and a more practical first-use experience. To try it, open the demo, choose a role, use fictional information only, and start with a simple legal-information question. The live demo and public source are linked above. I am proud of the multilingual framing, privacy guidance, and the effort to make the workspace more approachable for people who may not know where to begin.", "repo": "https://github.com/aaravkatiyar55-gif/legal-saathi", "demo": "https://legalsaathi-india.vercel.app/"}]...
|
[{"post_id": 51606, "ts": "202 [{"post_id": 51606, "ts": "2026-08-21T11:40:57Z", "hours": 0.5931, "text": "I’ve been refining Legal Saathi, an India-focused legal information and preparation workspace built to help people turn confusing situations into clearer legal questions, timelines, and next steps.This update focused on making the first-use experience more useful before a user even sends a message.What changed:Added a “Legal Preparation Compass” to help users identify the facts, timeline, documents, jurisdiction, and outcome that make a legal question clearer.Added a small interactive “Legal Clarity Sprint” with fictional examples. It helps users understand why context and a date-by-date timeline matter.Improved entry-screen copy and layout so users can choose actions such as asking a legal question, analysing a document, researching current law, reviewing an agreement, drafting a document, or preparing a case.Kept privacy and safety clear: Legal Saathi provides legal information and preparation support—it does not file cases, send notices, make payments, or replace an advocate.Improved Hindi and Hinglish support for the new onboarding content.Verified the production frontend build and deployed the update to the original Legal Saathi domain.Configured the currently available free AI model options for the supported Auto, Fast, and Flash choices, while keeping unavailable premium choices clearly marked instead of pretending they work.The goal is not to make legal help look flashy—it is to help someone arrive with a more understandable, safer, and better-organised question.I’m still treating real legal information carefully: users should avoid sharing unnecessary personal details, and urgent situations should be taken to the relevant authority, emergency service, or qualified advocate.Suggested screenshots to attach:Landing/workspace showing the Legal Preparation Compass. Legal Clarity Sprint question card. Action cards: Ask, Analyse, Research, Review, Draft, Prepare. The privacy/legal-information boundary and language controls."}, {"post_id": 51304, "ts": "2026-08-20T21:36:24Z", "hours": 1.6236, "text": "I have done some more updates in the app more privacy some rules and rewgulation also terms and conditions help section and settings section is now looking more good you can change theme by yourself and many things i can not tell all the features you can try out yourself and pls give me feedbacks And one more thing if anyone is signing in its totaly safe whether you sign in by email or google ok thanks"}, {"post_id": 50877, "ts": "2026-08-20T10:32:15Z", "hours": 14.6169, "text": "Legal Saathi — first build update. I am building an AI-assisted legal-information and preparation workspace for people who are more comfortable asking questions in English, Hindi, or Hinglish. I have been iterating on this project for around three months. I joined Stardance recently, so this is my first Legal Saathi devlog. I only just linked the existing tracker project to this Stardance project; the attached tracker image shows the project time currently recorded there. That tracker time is separate from any Stardance acceptance or payout decision. Right now I am working on guided legal workflows for asking questions, analysing documents, researching current-law questions, reviewing agreements, preparing editable drafts, organising case preparation, and comparing documents. Legal Saathi can help explain, organise, and prepare, but it does not file, send, pay, contact anyone, publish anything, or replace an advocate. The workspace is designed around English, Hindi, and Hinglish, with privacy guidance and clearer recovery after interrupted requests. The images show the linked tracker record plus current workspace flows in English and Hindi. This is still a work in progress, and I will only claim hosted features after I have directly verified them. I would love feedback on which workflow would be most useful to try first: document review, drafting a notice, a consumer issue, or preparing questions for an advocate."}]...
|
Title: Legal Saathi by @Aaravkatiyar | Stardance
Title: Legal Saathi by @Aaravkatiyar | Stardance
Devlogs:
Legal Saathi — first build update. I am building an AI-assisted legal-information and preparation workspace for people who are more comfortable asking questions in English, Hindi, or Hinglish. I have been iterating on this project for around three months. I joined Stardance recently, so this is my first Legal Saathi devlog. I only just linked the existing tracker project to this Stardance project; the attached tracker image shows the project time currently recorded there. That tracker time is separate from any Stardance acceptance or payout decision. Right now I am working on guided legal workflows for asking questions, analysing documents, researching current-law questions, reviewing agreements, preparing editable drafts, organising case preparation, and comparing documents. Legal Saathi can help explain, organise, and prepare, but it does not file, send, pay, contact anyone, publish anything, or replace an advocate. The workspace is designed around English, Hindi, and Hinglish, with privacy guidance and clearer recovery after interrupted requests. The images show the linked tracker record plus current workspace flows in English and Hindi. This is still a work in progress, and I will only claim hosted features after I have directly verified them. I would love feedback on which workflow would be most useful to try first: document review, drafting a notice, a consumer issue, or preparing questions for an advocate.
I have done some more updates in the app more privacy some rules and rewgulation also terms and conditions help section and settings section is now looking more good you can change theme by yourself and many things i can not tell all the features you can try out yourself and pls give me feedbacks And one more thing if anyone is signing in its totaly safe whether you sign in by email or google ok thanks
I’ve been refining Legal Saathi, an India-focused legal information and preparation workspace built to help people turn confusing situations into clearer legal questions, timelines, and next steps.This update focused on making the first-use experience more useful before a user even sends a message.What changed:Added a “Legal Preparation Compass” to help users identify the facts, timeline, documents, jurisdiction, and outcome that make a legal question clearer.Added a small interactive “Legal Clarity Sprint” with fictional examples. It helps users understand why context and a date-by-date timeline matter.Improved entry-screen copy and layout so users can choose actions such as asking a legal question, analysing a document, researching current law, reviewing an agreement, drafting a document, or preparing a case.Kept privacy and safety clear: Legal Saathi provides legal information and preparation support—it does not file cases, send notices, make payments, or replace an advocate.Improved Hindi and Hinglish support for the new onboarding content.Verified the production frontend build and deployed the update to the original Legal Saathi domain.Configured the currently available free AI model options for the supported Auto, Fast, and Flash choices, while keeping unavailable premium choices clearly marked instead of pretending they work.The goal is not to make legal help look flashy—it is to help someone arrive with a more understandable, safer, and better-organised question.I’m still treating real legal information carefully: users should avoid sharing unnecessary personal details, and urgent situations should be taken to the relevant authority, emergency service, or qualified advocate.Suggested screenshots to attach:Landing/workspace showing the Legal Preparation Compass. Legal Clarity Sprint question card. Action cards: Ask, Analyse, Research, Review, Draft, Prepare. The privacy/legal-information boundary and language controls.
Ships:
Legal Saathi is an AI-assisted legal-information and preparation workspace for English, Hindi, and Hinglish. It is designed to help people organise facts, understand safer next steps, review documents, and prepare draft structures — not to file, send, pay, contact anyone, or replace an advocate. I began iterating on this project before I knew about Stardance and joined the challenge recently, so the 15 hours shown here are the time currently recorded by the linked tracker. They are separate from the longer build journey. I have been building on limited hardware, which made testing and iteration slower but pushed me to focus on clearer flows, safer boundaries, and a more practical first-use experience. To try it, open the demo, choose a role, use fictional information only, and start with a simple legal-information question. The live demo and public source are linked above. I am proud of the multilingual framing, privacy guidance, and the effort to make the workspace more approachable for people who may not know where to begin....
|
1793541160.2327473
|
1793627560.2327473
|
|
50360
|
THE OBSERVER by @lakshay99_b | Stardance
|
lakshay99_b
|
1
|
2.27
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T09:44:00Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I finally shipped THE OBSERVER 🚀 I built The Observer as an astronomy-focused new-tab dashboard that turns a normal browser tab into a small digital observatory. I wanted it to feel alive, so I connected it with real space data like NASA’s Astronomy Picture of the Day, the ISS position, people currently in space, and upcoming launches. I also built the animated starfield, solar-system visualization, bookmarks, field notes, search, settings, and the rest of the interface myself using HTML, CSS and vanilla JavaScript. The most challenging part was getting all the different pieces to work together reliably. Working with multiple APIs, handling asynchronous data, dealing with loading/error states, creating the Canvas animations, and making the layout work across screen sizes taught me a lot more than I expected. I’m especially proud of the fact that this isn’t a one-click AI-generated website or a template that I simply customized. I spent the time designing the UI, writing the code, connecting the APIs, debugging things that broke, and iterating until it became something I was actually happy to ship. If you’re reviewing/testing it: open the deployed site and give it a moment to load the live data. Try the NASA picture, ISS tracker, people-in-space section, upcoming launch countdown, planet cards, search, bookmarks, settings, and Field Notes. You can also refresh the page to check that your saved settings and notes persist. This is the version I’m submitting for review, but I don’t consider it the end of The Observer. There are still plenty of things I’d love to improve and add. Thanks for checking out my project. 🌌 — Lakshay", "repo": "https://github.com/lakshay99b-dotcom/the-observer", "demo": "https://the-observer-angvey.vercel.app"}]...
|
[{"post_id": 51037, "ts": "202 [{"post_id": 51037, "ts": "2026-08-20T15:00:08Z", "hours": 0.2667, "text": "Spent some time cleaning up the top part of the page. Moved the search bar under the clock so it sits more like a normal new tab page. Made it wider too. Before it was stuck on the side and felt a bit awkward. Also shifted the GitHub / Stardance / YouTube buttons below the search bar and added simple icons to them. Looks cleaner now. Tried adding a space background image. First one I picked was too bright and made the text hard to read, so I darkened it with an overlay. It’s better now but I might still change the image later if I find a nicer one. Notes and name still save properly. APOD and ISS are working. Not a huge update, just small things that were bothering me. Feels a bit more usable now."}, {"post_id": 50859, "ts": "2026-08-20T09:43:10Z", "hours": 0.4308, "text": "JUST PLAYING THE FINAL CARD this is devlog#5 and now i am happy that this is finally at the stage of completenes."}, {"post_id": 50358, "ts": "2026-08-19T15:12:39Z", "hours": 0.3822, "text": "devlog#3 this might be going to be my last devlog for my this project. I FINALLY ADDED A VIDEO FETCHER FOR NASA’S VIDEO ADD IT IS ALMOST COMPLETE. FEATURES -search for anything -real time, date -nasa’s images and videoes -iss position tracker -launch tracker -name of humans currently in orbit although i know i was posting very less devlogs, but actually i am so stuck in code that i don’t remeber to do this."}, {"post_id": 50332, "ts": "2026-08-19T14:31:10Z", "hours": 0.4181, "text": "The Observer — Bringing the Solar System to Life 🌌The Observer is starting to feel like an actual space exploration tool.In this build, I’ve been working on the interactive solar-system visualization, where planets are displayed with their connections and can be explored to reveal information about them.The current version includes:🪐 Interactive planet visualization⭐ Animated background stars🌍 Planet information such as radius, gravity, mass and moons🛰️ Live ISS position data📍 ISS latitude, longitude, altitude and velocity💻 A completely browser-based interfaceThis screenshot is also a good reminder that development isn’t always clean 😅 — while building the star-field animation and resizing system, I ran into a JavaScript error in the console. I’m currently debugging it and improving the system so the visualization remains stable across different screen sizes.The goal for The Observer isn’t just to make something that looks like space.I want to turn it into a small interactive window into our actual universe — combining real astronomical data, visualization and exploration in one place.Still building. 🚀 #Stardance #HackClub #TheObserver #WebDevelopment #JavaScript #Space #Astronomy #BuildInPublic"}, {"post_id": 50170, "ts": "2026-08-19T08:52:18Z", "hours": 1.1272, "text": "this is my devlog#1 i today worked for nearly an hour messing with code and output. Currently the output in not attractive but i am just trying to improve it."}]...
|
Title: THE OBSERVER by @lakshay99_b | Stardance
D Title: THE OBSERVER by @lakshay99_b | Stardance
Devlogs:
this is my devlog#1 i today worked for nearly an hour messing with code and output. Currently the output in not attractive but i am just trying to improve it.
The Observer — Bringing the Solar System to Life 🌌The Observer is starting to feel like an actual space exploration tool.In this build, I’ve been working on the interactive solar-system visualization, where planets are displayed with their connections and can be explored to reveal information about them.The current version includes:🪐 Interactive planet visualization⭐ Animated background stars🌍 Planet information such as radius, gravity, mass and moons🛰️ Live ISS position data📍 ISS latitude, longitude, altitude and velocity💻 A completely browser-based interfaceThis screenshot is also a good reminder that development isn’t always clean 😅 — while building the star-field animation and resizing system, I ran into a JavaScript error in the console. I’m currently debugging it and improving the system so the visualization remains stable across different screen sizes.The goal for The Observer isn’t just to make something that looks like space.I want to turn it into a small interactive window into our actual universe — combining real astronomical data, visualization and exploration in one place.Still building. 🚀 #Stardance #HackClub #TheObserver #WebDevelopment #JavaScript #Space #Astronomy #BuildInPublic
devlog#3 this might be going to be my last devlog for my this project. I FINALLY ADDED A VIDEO FETCHER FOR NASA’S VIDEO ADD IT IS ALMOST COMPLETE. FEATURES -search for anything -real time, date -nasa’s images and videoes -iss position tracker -launch tracker -name of humans currently in orbit although i know i was posting very less devlogs, but actually i am so stuck in code that i don’t remeber to do this.
JUST PLAYING THE FINAL CARD this is devlog#5 and now i am happy that this is finally at the stage of completenes.
Spent some time cleaning up the top part of the page. Moved the search bar under the clock so it sits more like a normal new tab page. Made it wider too. Before it was stuck on the side and felt a bit awkward. Also shifted the GitHub / Stardance / YouTube buttons below the search bar and added simple icons to them. Looks cleaner now. Tried adding a space background image. First one I picked was too bright and made the text hard to read, so I darkened it with an overlay. It’s better now but I might still change the image later if I find a nicer one. Notes and name still save properly. APOD and ISS are working. Not a huge update, just small things that were bothering me. Feels a bit more usable now.
Ships:
I finally shipped THE OBSERVER 🚀 I built The Observer as an astronomy-focused new-tab dashboard that turns a normal browser tab into a small digital observatory. I wanted it to feel alive, so I connected it with real space data like NASA’s Astronomy Picture of the Day, the ISS position, people currently in space, and upcoming launches. I also built the animated starfield, solar-system visualization, bookmarks, field notes, search, settings, and the rest of the interface myself using HTML, CSS and vanilla JavaScript. The most challenging part was getting all the different pieces to work together reliably. Working with multiple APIs, handling asynchronous data, dealing with loading/error states, creating the Canvas animations, and making the layout work across screen sizes taught me a lot more than I expected. I’m especially proud of the fact that this isn’t a one-click AI-generated website or a template that I simply customized. I spent the time designing the UI, writing the code, connecting the APIs, debugging things that broke, and iterating until it became something I was actually happy to ship. If you’re reviewing/testing it: open the deployed site and give it a moment to load the live data. Try the NASA picture, ISS tracker, people-in-space section, upcoming launch countdown, planet cards, search, bookmarks, settings, and Field Notes. You can also refresh the page to check that your saved settings and notes persist. This is the version I’m submitting for review, but I don’t consider it the end of The Observer. There are still plenty of things I’d love to improve and add. Thanks for checking out my project. 🌌 — Lakshay...
|
1792878572.348227
|
1793742572.348227
|
|
51257
|
My own site! by @Ahnaafjim | Stardance
|
Ahnaafjim
|
1
|
0.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T09:15:51Z", "hours": 0.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "This is my first ever personal website, i made this using html and css and that is all i have knowledge on, i have very limited knowledge on javascript but since this was a project to learn something new i deided that i should probably try to use it. all the javascript that i used was made with the help of claude but everything else like the html and css was made by me, it actually took me a total of 4 hours but the thing is that hackatime did not assign the time for the first 3 hours and 40 minutes mainly because i did not have the project file open, iam currently seeing if they can resolve my issue but for now i’ve pretty much start with just a simple “hi welcome to my website” then i’ve build a big card with an image of me and a paragraph, i made it with Gemini but i made sure it was true to me Last i added the hobbies tab and that is when i realized it did not assign the hours so as a last ditch effort i added a new section thanking the user for clicking on my website just to meet the minimum coding time “quota” and that’s all.", "repo": "https://github.com/ahnaafjim/Final-personal-website", "demo": "https://ahnaafjim.github.io/Final-personal-website/"}]...
|
[{"post_id": 50839, "ts": "202 [{"post_id": 50839, "ts": "2026-08-20T08:47:47Z", "hours": 0.4544, "text": "made my first ever personal website!"}]...
|
Title: My own site! by @Ahnaafjim | Stardance
Dev Title: My own site! by @Ahnaafjim | Stardance
Devlogs:
made my first ever personal website!
Ships:
This is my first ever personal website, i made this using html and css and that is all i have knowledge on, i have very limited knowledge on javascript but since this was a project to learn something new i deided that i should probably try to use it. all the javascript that i used was made with the help of claude but everything else like the html and css was made by me, it actually took me a total of 4 hours but the thing is that hackatime did not assign the time for the first 3 hours and 40 minutes mainly because i did not have the project file open, iam currently seeing if they can resolve my issue but for now i’ve pretty much start with just a simple “hi welcome to my website” then i’ve build a big card with an image of me and a paragraph, i made it with Gemini but i made sure it was true to me Last i added the hobbies tab and that is when i realized it did not assign the hours so as a last ditch effort i added a new section thanking the user for clicking on my website just to meet the minimum coding time “quota” and that’s all....
|
1793480153.6001523
|
1793566553.6001523
|
|
51399
|
Focus Focus by @Afraz | Stardance
|
Afraz
|
1
|
2.0
|
3
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T09:12:27Z", "hours": 2.0, "multiplier": 1.82, "payout": 3, "blessing": "neutral", "text": "I made a Productivity OS. It works with HTML, CSS and JavaScript. The challenging parts were making the websites UI actually look good, if the UI looks bad, the user will have a really bad time, first the UI looked something that came out in the Windows ME era, then i slowly changed the UI to this one now. To use it, it has no hidden menus, no hamburger menues, it is just what you see when you open the site", "repo": "https://github.com/afrazakbar/Productivity-tracker/", "demo": "https://focus-focu.netlify.app/"}]...
|
[{"post_id": 50842, "ts": "202 [{"post_id": 50842, "ts": "2026-08-20T09:00:01Z", "hours": 1.5397, "text": "Added An Eisenhover Matrix for task optimization and pomodoro for better focus"}, {"post_id": 50338, "ts": "2026-08-19T14:38:36Z", "hours": 0.4047, "text": "Finished a Usable Version"}, {"post_id": 50292, "ts": "2026-08-19T13:24:06Z", "hours": 0.5392, "text": "Working on the css !"}]...
|
Title: Focus Focus by @Afraz | Stardance
Devlogs: Title: Focus Focus by @Afraz | Stardance
Devlogs:
Working on the css !
Finished a Usable Version
Added An Eisenhover Matrix for task optimization and pomodoro for better focus
Ships:
I made a Productivity OS. It works with HTML, CSS and JavaScript. The challenging parts were making the websites UI actually look good, if the UI looks bad, the user will have a really bad time, first the UI looked something that came out in the Windows ME era, then i slowly changed the UI to this one now. To use it, it has no hidden menus, no hamburger menues, it is just what you see when you open the site...
|
1793490634.6144376
|
1793577034.6144376
|
|
29287
|
le-fat-chaton by @unknown_lizzard | Stardance
|
unknown_lizzard
|
1
|
172.0
|
1818
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T09:10:04Z", "hours": 172.0, "multiplier": 14.88, "payout": 1818, "blessing": "neutral", "text": "at first I wanted to make a normal text model to try and learn more about llms, but I wanted to go further, agentic coding is a big thing so what better way to learn about llms than trying to make an agentic coding ai model. So thats how le gros chaton was born, the name originated from a joke on x about a fake huge mistral model coming out. to train this model I used already made datasets on hugging face, like fable5 traces, I also used kimi k3 as a teacher model to teach my smaller model to be good at coding and running commands, it kind of worked? but i’m not completely satisfied with the result, but at least i have learned some things and soon enough I will trying making an even better model! I dont have a demo url for you guys to try the model out since it was too expensive to run on a gpu so you’ll need to run it yourself, check it out on hugging face, follow the instructions and have fun.", "repo": "https://github.com/Mateooo93/le-gros-chaton", "demo": "https://mateooo93.github.io/le-gros-chaton/"}]...
|
[{"post_id": 50841, "ts": "202 [{"post_id": 50841, "ts": "2026-08-20T08:57:48Z", "hours": 3.2394, "text": "devlog #17 MI300X vLLM serving The old GPU box got destroyed, so I spun up a new MI300X with the same SSH key. ROCm 6.2 and Torch 2.5.1 were already set up, so getting the machine ready was easy. The hard part was vLLM. The pip version is CUDA only, and the ROCm Docker image only supported the multimodal version of Qwen3.5. Our model is text-only, so I had to patch the image. I added Qwen3_5ForCausalLM, fixed the hybrid KV-cache setup, added the missing Mamba state methods, and mapped the weight names so the text-only safetensors loaded correctly. I also patched the merged config to remove the M-RoPE and dtype fields. After that, the model served normally. I ran a one-task test with fix-git. The pipeline worked, but the model got stuck repeating git show until it ran out of turns. ##Shipped The merged model is now public on Hugging Face as mateo0093/le-gros-chaton under Apache-2.0. The 5×5 Terminal-Bench pilot finished at 3/25 (25%). All three successes were on fix-git, which shows that the SFT data was too focused on git tasks. I also tested RLVR with GRPO and a novelty bonus. The reward had almost no variance, so the training signal was basically useless. I saved the step-10 adapter but didn’t merge it. The GPU used about 12 hours of the 50 hour budget. All the vLLM patches and harness changes are in the repo. Main lesson: Qwen3.5’s hybrid model is a bit annoying to serve as text-only on ROCm. The model works but i think this was more of a learning opportunity, i know what i’ve done wrong and what i would have done differently, I will be working on a new model in a few weeks once I do a bit more research."}, {"post_id": 49761, "ts": "2026-08-18T18:01:12Z", "hours": 0.7992, "text": "Devlog #16, t4gpu The T4 16GB fought us the whole way. BF16 was baked into a pre quantized checkpoint, which made each step take 668 seconds because it had to use FP32 emulation. 2K context ran out of memory, the 248K logits ran out of memory too, FP16 started producing NaNs, Kaggle session limits locked us out, and the paged optimizer crashed last. Every fix seemed to create another problem. We eventually got it working with FP16 compute, chunked assistant token loss, autocast, and plain AdamW. Then we got access to an AMD MI300X with 192GB of memory. No quantization, 16K context, and enough memory to train full traces instead of cutting them off halfway through a message. Right now the 16K BF16 SFT run is training on the MI300X. Next up is merging the model, running Terminal Bench, and then RLVR."}, {"post_id": 49140, "ts": "2026-08-17T19:31:53Z", "hours": 16.24, "text": "Devlog 15 Traces are done The trace generator finally hit the target and went past it. We got 474 verified trajectories, zero malformed, zero duplicates, 19 templates, and a 95 percent self review pass rate. We were aiming for 450, so we ended up with 474. The dataset is synced to Hugging Face in two versions: raw and normalized. That part is finally done. It was the longest part of the project, and the model is only going to be as good as the traces, so getting this right was important The Kimi endpoint also stopped working. The old teacher account started returning 503 errors during generation and eventually died completely. We switched to a second Kimi account, which needed an extra header that we did not have at first. After figuring that out, generation resumed without losing any progress For GPUs, the 2070 still cannot realistically train the 9B. Kaggle hours are back, so we have 30 hours on a T4. The 4090 from my dad should arrive next week. The plan is to use the T4 for trajectory SFT, then the 4090 for merging, Terminal Bench evaluation, and RLVR. Status: 474 out of 474 traces done,"}, {"post_id": 42823, "ts": "2026-08-07T19:01:50Z", "hours": 8.9192, "text": "Devlog #14 worked a lot, still not done we worked a lot this week. honestly the main takeaway is that the fable5 sft was good but its traces alone weren’t enough the model needs way more verified agent trajectories, so we’ve been grinding those out (target is 450). we also ran out of credits mid-generation (the modal endpoint started failing with 503s) and had to switch to a second kimi account. took a minute to get the auth right the new endpoint wanted an extra header we didn’t have. after that the generator resumed fine and it’s been adding verified traces since (~340/450 now). biggest scare: we found out peft 0.20 silently drops the second adapter when you merge two at once. tested it and yeah, the outputs were identical with one or two adapters so our trajectory sft would have been lost in the merge. fixed it by merging sequentially (base first, then the second adapter). checked against the live model and it matches, so we’re good. my 2070 couldn’t fit the 9b model for real training, so we’re waiting on a 4090 (my dad’s) next week. all the scripts are ready sft, merge, rlvr, eval, even a setup script for the gpu box. just need the hardware. status: 340/450 traces, everything ready, waiting on the 4090. we’ve worked a lot, but the real run hasn’t happened yet :/."}, {"post_id": 41470, "ts": "2026-08-05T18:10:35Z", "hours": 18.5394, "text": "Devlog #13 we worked ALOT so this was the grind day. 18 hours of training babysitting and honestly it felt like 40. the modal training kept dying every hour. the client would crash and take the whole run with it. we restarted like 3 times before figuring out modal run -d (detached) keeps it alive on their servers even if our computer shits the bed. tested it by literally killing the process and it kept going. ez fix but we worked ALOT to get there. then we ran out of credits at 93% and switched accounts, but the other account “finished” the run in 6 MILLISECONDS which was sus af. turns out it just re-uploaded the old checkpoint and trained nothing. the resume logic had a bug where it thought 6200 >= 3799 meant done. patched it. another lesson learned. also we made the model creative lol. added diversity sampling to the trajectory generator (5 solutions per task, keep the novel ones), a novelty bonus reward for RLVR, and a creativity rule in the self-awareness prompt. so the fat cat will be innovative, not just memorize. current state: SFT is done at 91% (good enough), resume bug fixed, phase 2 (trajectory SFT with creativity + self-awareness baked in) is up next. 18 hours. we worked ALOT. the fat cat will be worth it 😼"}, {"post_id": 40883, "ts": "2026-08-04T18:02:57Z", "hours": 10.0019, "text": "Devlog #12 training kept dying for the last 10 hours the training kept crashing. not the model, not the gp but the modal client. every hour or so it would lose connection and take the whole run down with it. we’d restart, it’d run an hour, crash again. three times. the fix was simple: modal run -d runs detached, so the training lives on modal’s servers and doesn’t care if our local process dies. killed the client on purpose to test — training kept going. that’s it, that’s the fix. also worried the loss wasn’t moving but it was fine. lr had wound down on the crashed run, looked scary, re-warmed fine after restart. loss is 0.668 now and dropping. status: 72% through the fable5 sft. checkpoints safe on hf. after this: trajectory sft with the self-awareness stuff baked in, then rlvr, then real numbers."}, {"post_id": 40601, "ts": "2026-08-04T08:01:50Z", "hours": 11.4342, "text": "** Devlog #11 half way there! ** the training is going pretty well, I fixed a few bugs with the code and added documentation training is at 50%!"}, {"post_id": 40369, "ts": "2026-08-03T20:35:48Z", "hours": 8.8192, "text": "** Devlog #10 the full training is finally running ** ok so phase 1 full training is officially going. 160k rows of fable5 data on modal, resuming from the 12.8k rows we got on kaggle before the session cap killed us. we’re at ~27% and it should take like a day and a half total. big stuff that happened since last time: fast kernels actually work now. the model has hybrid linear-attention layers and transformers kept falling back to a slow torch path. the fix was: nvidia cuda-devel base image (needs nvcc to compile), installing flash-linear-attention[cuda] + causal-conv1d , and pinning torch 2.10. took a couple tries but now training is like 4x faster than it would be without them. had to split across two modal accounts. first account ran out of credits, so we set up a second profile and the script auto-resumes from the huggingface checkpoints when it dies. it’s been rock solid, it happens every 20% plus every hour it saves and uploads. lost zero progress to the account switch. we gave the model a personality. it’s called Le Gros Chaton (the fat cat) and we’re training self-awareness into it, not as a system prompt but baked into the weights. state-sheets ([STATE] goal/known/tried/failed/next), metacognition lines, and self-review at the end of tasks. the trajectory sft masks out the system prompt so the model learns to track its own state as behavior, not as a script it repeats. did some benchmark research too. base qwen3.5-9b scores 9.2% on terminal-bench 2.0. comparable 9b fine-tunes hit 24-28%. that’s our realistic target which is 2.5-3x the baseline. honestly at this point it’s just… waiting. the pipeline works, the checkpoints work, the account switching works. now the model just needs to finish cooking. then trajectory sft, then rlvr, then we actually see the benchmark numbers."}, {"post_id": 40029, "ts": "2026-08-03T11:45:18Z", "hours": 13.3375, "text": "# Devlog #9 — finally training, and now with checkpoints! so last time i was stuck debugging phase 1/2 on kaggle. well, still debugging lol, but we actually got somewhere. the first issue was transformers not knowing what the qwen3.5 model was, had to pin transformers 5.14.1. then tokenizers kept fighting with it (0.22.0 wants old huggingface-hub, 0.22.1 is fine). then torch on kaggle kept giving a p100 which modern pytorch just refuses to run on, so we had to tell it to give us a t4 instead. then the fun one: out of memory on the very first step. batch 2 was too big for the gpu they gave us, so batch 1 it is. also i learned the hard way that kaggle kills your session after 12 hours. the first long run got cancelled at 80% done and i had nothing saved. so now the training script saves a checkpoint every 20% + every hour and uploads it to huggingface, so if it dies we just resume. it actually saved us, the run got cut and we picked right back up from the checkpoint. right now the full 160k sft is running on modal (paying for gpus now, kaggle time ran out). it’s at like 9% and should take a couple days. we’ll see how it goes, and then rlvr after that. the end of the errors… might actually be in sight this time."}, {"post_id": 38292, "ts": "2026-07-31T16:38:26Z", "hours": 5.4886, "text": "Devlog #8 — phase 2 fixes I’ve been getting a lot of issues with the first two for this model. I’ve gotten a lot of errors in Kaggle and Colab. I’ve tried everything, and I’m still fixing the code. It’s taking a good amount of time, but I think I’m almost there. I’ve gone through a lot of errors, and I’m probably right at the end of the errors. I might be able to start actually training it tomorrow, but for now I’ll stop."}, {"post_id": 38079, "ts": "2026-07-31T09:21:05Z", "hours": 10.3872, "text": "Devlog #7 — phase 2 running phase 1 (sft) is done and phase 2 (rlvr) is running now on kaggle. this is the part where the model actually gets trained on solving coding problems with our verifier giving rewards, not just imitating fable data. honestly this is the part i care about more. sft just makes it mimic. rlvr is where it learns to actually fix code. grpo with proportional rewards = the model gets partial credit for getting some tests passing, which is way better than all-or-nothing. while it runs im just waiting tbh. not much to build. the pipeline works, the eval works, the agent works. next after this: full benchmark run. humaneval, swebench, agentic eval. we’ll see if all this actually made the model good at coding or if i just spent a week on a worse qwen lol."}, {"post_id": 37100, "ts": "2026-07-29T16:05:47Z", "hours": 8.4747, "text": "Devlog #6 — training done (phase 1) finally got the qwen finetune to actually run with 625 steps, took like 8 hours on kaggle. it finished. havent run full benchmarks yet cause i was busy building other stuff while it trained. but it trained. thats the main thing. I also built a swebench agent that can actually navigate repos and edit files and make patches. not just the basic agent loop that runs commands. this one actually works for real se tasks. also added swebench eval so we can measure if the model is actually good at fixing bugs and coding then i’ll run the benchmarks on the trained model and see if its actually any good lol."}, {"post_id": 36367, "ts": "2026-07-28T13:16:16Z", "hours": 13.6447, "text": "Devlog #5 — ok so I’m not building a 10B MoE anymore So last time I had this whole plan. 10B MoE, from scratch, SwiGLU, GQA, all that. And I actually built it. Like 60 Python files, tests passing, RL pipeline, self-play data gen, the whole thing. It was kinda sick ngl. But then I was like… wait. I’m building an ENGINE that TRAINS coding agents. Not the agent. And the engine is cool but training a 10B model from zero on $30 of Modal credits is not how you beat GLM-5.2. You know what beats GLM-5.2? Fine-tuning a model that already knows how to code. the pivot (again) Qwen3.5-9B came out. 9B params. Fits on a Kaggle T4 with 4-bit QLoRA. And Kaggle gives me 30 HOURS of free GPU. 2× T4. That’s literally free training. So I had to adapt the entire codebase to work with HuggingFace models instead of my custom GPT. Made like 5 new files. All the verifier/agent/reward stuff I built for the MoE? Now works with Qwen too. Didn’t have to rewrite everything — just slapped a Qwen adapter on top. the dataset Found this dataset on HF: Nexlab/fable5-agentic-coding-sft . 160K rows of Claude Fable 5 output. Like actual coding traces — building games, fixing bugs, writing shell scripts. Not just “write a function that passes a test.” Real agentic stuff. Training right now actually. Qwen3.5-9B on 10K rows. Should take like 8 hours on 2× T4. We’ll see if it doesn’t OOM again lol. While waiting for training I read papers: Routing-Free MoE — what if experts just decide for themselves when to activate? No router, no top-k. Already implemented it in model.py. GLM-5.2 has this IndexShare thing that reuses computation across layers. 2.9× cheaper at 1M context. Maybe later. DeepSWE trained Qwen3-32B with RL only and got 42% SWE-Bench. So our approach is valid. the first phase of training is going on right now, i’ll post another devlog once it’s done!"}, {"post_id": 20759, "ts": "2026-07-02T13:35:48Z", "hours": 5.6175, "text": "Devlog #4 — wait, why am I training on Wikipedia? Last week I got le gros chaton (my from-scratch transformer) to actually train. Ran it on wikitext-103 in a Colab notebook on a free T4, 12k steps, val loss hit 3.73. Felt insane for something I wrote myself from PyTorch primitives. But then I looked at what it generated. It could do English-ish. It knew about Singapore and stuff. The moment I asked it to write code? Total garbage. Fair enough. It has never seen a single line of code. the pivot I thought about what I actually want this thing to do. Answer: a coding agent. Something that mogs on terminal-bench, that you give a task and it runs commands, fixes its own bugs, gets it done. Qwen-coder territory. Wikitext won’t get me there. You can’t learn Python from Wikipedia articles about the snake. So the plan changed. The big project is now le fat chaton, the bigger sibling. And it’s a coder. the architecture stuff I went down a rabbit hole on what frontier coding models do. They’re basically all mixture-of-experts (MoE) now. Qwen3-coder is 480B params / 35B active. Deepseek-coder-v2-lite is 16B / 2.4B active. The trick: big total params (knows a lot) small active params per token (runs fast and cheap) Runs like a small model, knows like a big one. Exactly what I want for a snappy terminal agent. So I rewrote the model. Added an MoE class. A gate routes each token to its top-2 experts out of 8, with a load-balance loss so the router doesn’t pick the same expert every time. Also threw in: SwiGLU — gated MLP, basically free quality, every Llama uses it GQA — smaller KV cache, faster long-context decoding (the agent harness needs this) Shared expert — DeepSeek-style, one always-on expert for common knowledge Fat config math: ~10.25B total / ~3.65B active. Not building that locally, it would crash my PC again. the data question Wikitext was a fine proof. For the coder I need code. Options: smollm-corpus python-edu — 7.6M educational Python files, high quality, but gated behind S3 (slow pull) starcoderdata python — ~50GB Python, streams fast from HF, the Codellama workhorse the-stack-v2 — 900B tokens, overkill for my budget Going with starcoderdata Python, blended with ~15-20% cosmopedia (prose) so it has general knowledge. That blend is the Qwen-coder recipe. am I warm-starting from wikitext? No. Different architecture entirely (17M dense vs 10B MoE, weights don’t fit). Even if they did, warm-starting a coder from prose wastes compute unlearning prose. Fresh random init on the mixed corpus. Clean slate. This went from “lets make a little model that talks” to “lets build a coding agent that mogs the big labs” real fast. Ambitious for a solo student with a 2070 and $30 of Modal credits? Absolutely. Doing it anyway? Yeah. Next: the eval harness so I can measure if this is making the model better at code (pass@1, pass@5 on HumanEval). Val loss going down means nothing if it still can’t write a"}, {"post_id": 20019, "ts": "2026-07-01T14:54:30Z", "hours": 0.3917, "text": "DEVLOG #3: teaching it to talk (fine-tuning) the base model was a completer. ask it “what is 2+2?” and it just continues the sentence. fine-tuning is what makes it actually answer: same model, more training, but on prompt/response pairs. i wrote ~48 Q&A pairs by hand. trivia, math, jokes, hellos. all in the same rigid Q: …\\nA: format. new trick: loss masking. only the response counts toward the loss, not the prompt, so i set prompt targets to -100 (pytorch’s cross_entropy ignores that automatically, zero model changes). tested the mask boundary in isolation first. wrong mask = 400 steps of training the wrong thing. the run: load base weights, 400 steps, lower lr. loss went 12.3 to 0.02.then the reality check. it nails everything in the set. capitals, math, jokes, “who are you,” verbatim. but something NOT in the set? slop. 17M params + 48 examples is a lookup table, not a reasoner. real chat models are billions of params. not a bug, just the scale.but it’s finally coherent. the first base model was word salad. now you type a question and get a real answer back that makes sense. it’s tiny and dumb and can’t generalize, but it actually talks. that’s the win. full pipeline done: tokenizer, model from scratch, pretrain on wikipedia, fine-tune, chat. mine, it talks, i understand every piece."}, {"post_id": 20006, "ts": "2026-07-01T14:16:25Z", "hours": 0.4139, "text": "DEVLOG #2: real data, real loss, and a lesson about memorizing split my data into 90% train / 10% val. before i only had one number, training loss, and it lies to you. you can’t tell learning from memorizing with just that. now i check both every 200 steps. rule: both go down = learning. train down but val up = memorizing = useless.also made the model bigger, 7M -> 17.7M params. immediately ran out of gpu memory lol. dropped batch size to 32 to fit my 8GB card.training: 3000 steps. loss started at 158 (should be ~10 for random , my first pass blew up and made garbage). recovered fast though, didn’t ruin the run. right now the llm is spitting out gibberish and random words but after finetuning it should sound more coherent"}, {"post_id": 19989, "ts": "2026-07-01T13:48:37Z", "hours": 36.31, "text": "DEVLOG #1: wiki text training so i finally got the model to train on actual data instead of that one paragraph about singapore nature reserves lol. before this the whole thing was basically a joke, i trained a 7M parameter transformer on like 134 tokens and it just spat out jumbled fragments of the same sentence over and over. cool proof it worked but definitely not a “language model” in any real sense.anyway. the big move this week was swapping out the tiny json file for wikitext-2. that’s a huggingface dataset of wikipedia text, about 2.4 million tokens once you filter out the empty lines. that’s… a lot more than 134 lol. encoding it took maybe half a minute with tiktoken and the whole tensor fits on my gpu no problem (it’s only like 8MB of ints)."}]...
|
Title: le-fat-chaton by @unknown_lizzard | Stardan Title: le-fat-chaton by @unknown_lizzard | Stardance
Devlogs:
DEVLOG #1: wiki text training so i finally got the model to train on actual data instead of that one paragraph about singapore nature reserves lol. before this the whole thing was basically a joke, i trained a 7M parameter transformer on like 134 tokens and it just spat out jumbled fragments of the same sentence over and over. cool proof it worked but definitely not a “language model” in any real sense.anyway. the big move this week was swapping out the tiny json file for wikitext-2. that’s a huggingface dataset of wikipedia text, about 2.4 million tokens once you filter out the empty lines. that’s… a lot more than 134 lol. encoding it took maybe half a minute with tiktoken and the whole tensor fits on my gpu no problem (it’s only like 8MB of ints).
DEVLOG #2: real data, real loss, and a lesson about memorizing split my data into 90% train / 10% val. before i only had one number, training loss, and it lies to you. you can’t tell learning from memorizing with just that. now i check both every 200 steps. rule: both go down = learning. train down but val up = memorizing = useless.also made the model bigger, 7M -> 17.7M params. immediately ran out of gpu memory lol. dropped batch size to 32 to fit my 8GB card.training: 3000 steps. loss started at 158 (should be ~10 for random , my first pass blew up and made garbage). recovered fast though, didn’t ruin the run. right now the llm is spitting out gibberish and random words but after finetuning it should sound more coherent
DEVLOG #3: teaching it to talk (fine-tuning) the base model was a completer. ask it “what is 2+2?” and it just continues the sentence. fine-tuning is what makes it actually answer: same model, more training, but on prompt/response pairs. i wrote ~48 Q&A pairs by hand. trivia, math, jokes, hellos. all in the same rigid Q: …\nA: format. new trick: loss masking. only the response counts toward the loss, not the prompt, so i set prompt targets to -100 (pytorch’s cross_entropy ignores that automatically, zero model changes). tested the mask boundary in isolation first. wrong mask = 400 steps of training the wrong thing. the run: load base weights, 400 steps, lower lr. loss went 12.3 to 0.02.then the reality check. it nails everything in the set. capitals, math, jokes, “who are you,” verbatim. but something NOT in the set? slop. 17M params + 48 examples is a lookup table, not a reasoner. real chat models are billions of params. not a bug, just the scale.but it’s finally coherent. the first base model was word salad. now you type a question and get a real answer back that makes sense. it’s tiny and dumb and can’t generalize, but it actually talks. that’s the win. full pipeline done: tokenizer, model from scratch, pretrain on wikipedia, fine-tune, chat. mine, it talks, i understand every piece.
Devlog #4 — wait, why am I training on Wikipedia? Last week I got le gros chaton (my from-scratch transformer) to actually train. Ran it on wikitext-103 in a Colab notebook on a free T4, 12k steps, val loss hit 3.73. Felt insane for something I wrote myself from PyTorch primitives. But then I looked at what it generated. It could do English-ish. It knew about Singapore and stuff. The moment I asked it to write code? Total garbage. Fair enough. It has never seen a single line of code. the pivot I thought about what I actually want this thing to do. Answer: a coding agent. Something that mogs on terminal-bench, that you give a task and it runs commands, fixes its own bugs, gets it done. Qwen-coder territory. Wikitext won’t get me there. You can’t learn Python from Wikipedia articles about the snake. So the plan changed. The big project is now le fat chaton, the bigger sibling. And it’s a coder. the architecture stuff I went down a rabbit hole on what frontier coding models do. They’re basically all mixture-of-experts (MoE) now. Qwen3-coder is 480B params / 35B active. Deepseek-coder-v2-lite is 16B / 2.4B active. The trick: big total params (knows a lot) small active params per token (runs fast and cheap) Runs like a small model, knows like a big one. Exactly what I want for a snappy terminal agent. So I rewrote the model. Added an MoE class. A gate routes each token to its top-2 experts out of 8, with a load-balance loss so the router doesn’t pick the same expert every time. Also threw in: SwiGLU — gated MLP, basically free quality, every Llama uses it GQA — smaller KV cache, faster long-context decoding (the agent harness needs this) Shared expert — DeepSeek-style, one always-on expert for common knowledge Fat config math: ~10.25B total / ~3.65B active. Not building that locally, it would crash my PC again. the data question Wikitext was a fine proof. For the coder I need code. Options: smollm-corpus python-edu — 7.6M educational Python files, high quality, but gated behind S3 (slow pull) starcoderdata python — ~50GB Python, streams fast from HF, the Codellama workhorse the-stack-v2 — 900B tokens, overkill for my budget Going with starcoderdata Python, blended with ~15-20% cosmopedia (prose) so it has general knowledge. That blend is the Qwen-coder recipe. am I warm-starting from wikitext? No. Different architecture entirely (17M dense vs 10B MoE, weights don’t fit). Even if they did, warm-starting a coder from prose wastes compute unlearning prose. Fresh random init on the mixed corpus. Clean slate. This went from “lets make a little model that talks” to “lets build a coding agent that mogs the big labs” real fast. Ambitious for a solo student with a 2070 and $30 of Modal credits? Absolutely. Doing it anyway? Yeah. Next: the eval harness so I can measure if this is making the model better at code (pass@1, pass@5 on HumanEval). Val loss going down means nothing if it still can’t write a
Devlog #5 — ok so I’m not building a 10B MoE anymore So last time I had this whole plan. 10B MoE, from scratch, SwiGLU, GQA, all that. And I actually built it. Like 60 Python files, tests passing, RL pipeline, self-play data gen, the whole thing. It was kinda sick ngl. But then I was like… wait. I’m building an ENGINE that TRAINS coding agents. Not the agent. And the engine is cool but training a 10B model from zero on $30 of Modal credits is not how you beat GLM-5.2. You know what beats GLM-5.2? Fine-tuning a model that already knows how to code. the pivot (again) Qwen3.5-9B came out. 9B params. Fits on a Kaggle T4 with 4-bit QLoRA. And Kaggle gives me 30 HOURS of free GPU. 2× T4. That’s literally free training. So I had to adapt the entire codebase to work with HuggingFace models instead of my custom GPT. Made like 5 new files. All the verifier/agent/reward stuff I built for the MoE? Now works with Qwen too. Didn’t have to rewrite everything — just slapped a Qwen adapter on top. the dataset Found this dataset on HF: Nexlab/fable5-agentic-coding-sft . 160K rows of Claude Fable 5 output. Like actual coding traces — building games, fixing bugs, writing shell scripts. Not just “write a function that passes a test.” Real agentic stuff. Training right now actually. Qwen3.5-9B on 10K rows. Should take like 8 hours on 2× T4. We’ll see if it doesn’t OOM again lol. While waiting for training I read papers: Routing-Free MoE — what if experts just decide for themselves when to activate? No router, no top-k. Already implemented it in model.py. GLM-5.2 has this IndexShare thing that reuses computation across layers. 2.9× cheaper at 1M context. Maybe later. DeepSWE trained Qwen3-32B with RL only and got 42% SWE-Bench. So our approach is valid. the first phase of training is going on right now, i’ll post another devlog once it’s done!
Devlog #6 — training done (phase 1) finally got the qwen finetune to actually run with 625 steps, took like 8 hours on kaggle. it finished. havent run full benchmarks yet cause i was busy building other stuff while it trained. but it trained. thats the main thing. I also built a swebench agent that can actually navigate repos and edit files and make patches. not just the basic agent loop that runs commands. this one actually works for real se tasks. also added swebench eval so we can measure if the model is actually good at fixing bugs and coding then i’ll run the benchmarks on the trained model and see if its actually any good lol.
Devlog #7 — phase 2 running phase 1 (sft) is done and phase 2 (rlvr) is running now on kaggle. this is the part where the model actually gets trained on solving coding problems with our verifier giving rewards, not just imitating fable data. honestly this is the part i care about more. sft just makes it mimic. rlvr is where it learns to actually fix code. grpo with proportional rewards = the model gets partial credit for getting some tests passing, which is way better than all-or-nothing. while it runs im just waiting tbh. not much to build. the pipeline works, the eval works, the agent works. next after this: full benchmark run. humaneval, swebench, agentic eval. we’ll see if all this actually made the model good at coding or if i just spent a week on a worse qwen lol.
Devlog #8 — phase 2 fixes I’ve been getting a lot of issues with the first two for this model. I’ve gotten a lot of errors in Kaggle and Colab. I’ve tried everything, and I’m still fixing the code. It’s taking a good amount of time, but I think I’m almost there. I’ve gone through a lot of errors, and I’m probably right at the end of the errors. I might be able to start actually training it tomorrow, but for now I’ll stop.
# Devlog #9 — finally training, and now with checkpoints! so last time i was stuck debugging phase 1/2 on kaggle. well, still debugging lol, but we actually got somewhere. the first issue was transformers not knowing what the qwen3.5 model was, had to pin transformers 5.14.1. then tokenizers kept fighting with it (0.22.0 wants old huggingface-hub, 0.22.1 is fine). then torch on kaggle kept giving a p100 which modern pytorch just refuses to run on, so we had to tell it to give us a t4 instead. then the fun one: out of memory on the very first step. batch 2 was too big for the gpu they gave us, so batch 1 it is. also i learned the hard way that kaggle kills your session after 12 hours. the first long run got cancelled at 80% done and i had nothing saved. so now the training script saves a checkpoint every 20% + every hour and uploads it to huggingface, so if it dies we just resume. it actually saved us, the run got cut and we picked right back up from the checkpoint. right now the full 160k sft is running on modal (paying for gpus now, kaggle time ran out). it’s at like 9% and should take a couple days. we’ll see how it goes, and then rlvr after that. the end of the errors… might actually be in sight this time.
** Devlog #10 the full training is finally running ** ok so phase 1 full training is officially going. 160k rows of fable5 data on modal, resuming from the 12.8k rows we got on kaggle before the session cap killed us. we’re at ~27% and it should take like a day and a half total. big stuff that happened since last time: fast kernels actually work now. the model has hybrid linear-attention layers and transformers kept falling back to a slow torch path. the fix was: nvidia cuda-devel base image (needs nvcc to compile), installing flash-linear-attention[cuda] + causal-conv1d , and pinning torch 2.10. took a couple tries but now training is like 4x faster than it would be without them. had to split across two modal accounts. first account ran out of credits, so we set up a second profile and the script auto-resumes from the huggingface checkpoints when it dies. it’s been rock solid, it happens every 20% plus every hour it saves and uploads. lost zero progress to the account switch. we gave the model a personality. it’s called Le Gros Chaton (the fat cat) and we’re training self-awareness into it, not as a system prompt but baked into the weights. state-sheets ([STATE] goal/known/tried/failed/next), metacognition lines, and self-review at the end of tasks. the trajectory sft masks out the system prompt so the model learns to track its own state as behavior, not as a script it repeats. did some benchmark research too. base qwen3.5-9b scores 9.2% on terminal-bench 2.0. comparable 9b fine-tunes hit 24-28%. that’s our realistic target which is 2.5-3x the baseline. honestly at this point it’s just… waiting. the pipeline works, the checkpoints work, the account switching works. now the model just needs to finish cooking. then trajectory sft, then rlvr, then we actually see the benchmark numbers.
** Devlog #11 half way there! ** the training is going pretty well, I fixed a few bugs with the code and added documentation training is at 50%!
Devlog #12 training kept dying for the last 10 hours the training kept crashing. not the model, not the gp but the modal client. every hour or so it would lose connection and take the whole run down with it. we’d restart, it’d run an hour, crash again. three times. the fix was simple: modal run -d runs detached, so the training lives on modal’s servers and doesn’t care if our local process dies. killed the client on purpose to test — training kept going. that’s it, that’s the fix. also worried the loss wasn’t moving but it was fine. lr had wound down on the crashed run, looked scary, re-warmed fine after restart. loss is 0.668 now and dropping. status: 72% through the fable5 sft. checkpoints safe on hf. after this: trajectory sft with the self-awareness stuff baked in, then rlvr, then real numbers.
Devlog #13 we worked ALOT so this was the grind day. 18 hours of training babysitting and honestly it felt like 40. the modal training kept dying every hour. the client would crash and take the whole run with it. we restarted like 3 times before figuring out modal run -d (detached) keeps it alive on their servers even if our computer shits the bed. tested it by literally killing the process and it kept going. ez fix but we worked ALOT to get there. then we ran out of credits at 93% and switched accounts, but the other account “finished” the run in 6 MILLISECONDS which was sus af. turns out it just re-uploaded the old checkpoint and trained nothing. the resume logic had a bug where it thought 6200 >= 3799 meant done. patched it. another lesson learned. also we made the model creative lol. added diversity sampling to the trajectory generator (5 solutions per task, keep the novel ones), a novelty bonus reward for RLVR, and a creativity rule in the self-awareness prompt. so the fat cat will be innovative, not just memorize. current state: SFT is done at 91% (good enough), resume bug fixed, phase 2 (trajectory SFT with creativity + self-awareness baked in) is up next. 18 hours. we worked ALOT. the fat cat will be worth it 😼
Devlog #14 worked a lot, still not done we worked a lot this week. honestly the main takeaway is that the fable5 sft was good but its traces alone weren’t enough the model needs way more verified agent trajectories, so we’ve been grinding those out (target is 450). we also ran out of credits mid-generation (the modal endpoint started failing with 503s) and had to switch to a second kimi account. took a minute to get the auth right the new endpoint wanted an extra header we didn’t have. after that the generator resumed fine and it’s been adding verified traces since (~340/450 now). biggest scare: we found out peft 0.20 silently drops the second adapter when you merge two at once. tested it and yeah, the outputs were identical with one or two adapters so our trajectory sft would have been lost in the merge. fixed it by merging sequentially (base first, then the second adapter). checked against the live model and it matches, so we’re good. my 2070 couldn’t fit the 9b model for real training, so we’re waiting on a 4090 (my dad’s) next week. all the scripts are ready sft, merge, rlvr, eval, even a setup script for the gpu box. just need the hardware. status: 340/450 traces, everything ready, waiting on the 4090. we’ve worked a lot, but the real run hasn’t happened yet :/.
Devlog 15 Traces are done The trace generator finally hit the target and went past it. We got 474 verified trajectories, zero malformed, zero duplicates, 19 templates, and a 95 percent self review pass rate. We were aiming for 450, so we ended up with 474. The dataset is synced to Hugging Face in two versions: raw and normalized. That part is finally done. It was the longest part of the project, and the model is only going to be as good as the traces, so getting this right was important The Kimi endpoint also stopped working. The old teacher account started returning 503 errors during generation and eventually died completely. We switched to a second Kimi account, which needed an extra header that we did not have at first. After figuring that out, generation resumed without losing any progress For GPUs, the 2070 still cannot realistically train the 9B. Kaggle hours are back, so we have 30 hours on a T4. The 4090 from my dad should arrive next week. The plan is to use the T4 for trajectory SFT, then the 4090 for merging, Terminal Bench evaluation, and RLVR. Status: 474 out of 474 traces done,
Devlog #16, t4gpu The T4 16GB fought us the whole way. BF16 was baked into a pre quantized checkpoint, which made each step take 668 seconds because it had to use FP32 emulation. 2K context ran out of memory, the 248K logits ran out of memory too, FP16 started producing NaNs, Kaggle session limits locked us out, and the paged optimizer crashed last. Every fix seemed to create another problem. We eventually got it working with FP16 compute, chunked assistant token loss, autocast, and plain AdamW. Then we got access to an AMD MI300X with 192GB of memory. No quantization, 16K context, and enough memory to train full traces instead of cutting them off halfway through a message. Right now the 16K BF16 SFT run is training on the MI300X. Next up is merging the model, running Terminal Bench, and then RLVR.
devlog #17 MI300X vLLM serving The old GPU box got destroyed, so I spun up a new MI300X with the same SSH key. ROCm 6.2 and Torch 2.5.1 were already set up, so getting the machine ready was easy. The hard part was vLLM. The pip version is CUDA only, and the ROCm Docker image only supported the multimodal version of Qwen3.5. Our model is text-only, so I had to patch the image. I added Qwen3_5ForCausalLM, fixed the hybrid KV-cache setup, added the missing Mamba state methods, and mapped the weight names so the text-only safetensors loaded correctly. I also patched the merged config to remove the M-RoPE and dtype fields. After that, the model served normally. I ran a one-task test with fix-git. The pipeline worked, but the model got stuck repeating git show until it ran out of turns. ##Shipped The merged model is now public on Hugging Face as mateo0093/le-gros-chaton under Apache-2.0. The 5×5 Terminal-Bench pilot finished at 3/25 (25%). All three successes were on fix-git, which shows that the SFT data was too focused on git tasks. I also tested RLVR with GRPO and a novelty bonus. The reward had almost no variance, so the training signal was basically useless. I saved the step-10 adapter but didn’t merge it. The GPU used about 12 hours of the 50 hour budget. All the vLLM patches and harness changes are in the repo. Main lesson: Qwen3.5’s hybrid model is a bit annoying to serve as text-only on ROCm. The model works but i think this was more of a learning opportunity, i know what i’ve done wrong and what i would have done differently, I will be working on a new model in a few weeks once I do a bit more research.
Ships:
at first I wanted to make a normal text model to try and learn more about llms, but I wanted to go further, agentic coding is a big thing so what better way to learn about llms than trying to make an agentic coding ai model. So thats how le gros chaton was born, the name originated from a joke on x about a fake huge mistral model coming out. to train this model I used already made datasets on hugging face, like fable5 traces, I also used kimi k3 as a teacher model to teach my smaller model to be good at coding and running commands, it kind of worked? but i’m not completely satisfied with the result, but at least i have learned some things and soon enough I will trying making an even better model! I dont have a demo url for you guys to try the model out since it was too expensive to run on a gpu so you’ll need to run it yourself, check it out on hugging face, follow the instructions and have fun....
|
1792878632.7358682
|
1793742632.7358682
|
|
27450
|
JBootloader by @jphwb | Stardance
|
jphwb
|
1
|
15.0
|
204
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T06:38:10Z", "hours": 15.0, "multiplier": 13.86, "payout": 204, "blessing": "neutral", "text": "JBootloader is a custom bootloader and kernel with a shell that has basic functionality made in c and assembly x86_64. The kernel currently can load a fat-16 filesystem image in read only mode and commands such as cat and ls can be used. Everything that is required to boot is all compiled into os.bin (Releases) including the kernel, stage1/stage2.asm and the fat16 filesystem which is designed to work on QEMU. This project isnt made to be actually used (obviously) but its more of a learning journey helping me learn how a bootloader and kernel work. Initially I wanted to just make a bootloader like GRUB that could boot any operating system but at the time I was overestimating my skills in asm so I decided to steer off and make a simple kernel instead. The challenging part was the agonizing “why won’t the fat-16 partition show up” because it just would show up no matter what I did so I pulled up gemini and put in my fat_16.c code and asked it to edit my code to push out any debugging info to the vm so that I’d be able to see what went wrong, using this I overcame my problems. Also at the start I spent almost an hour trying to find an issue since my C/C++ extension in VS wasn’t working and my build command just refused to tell me what was wrong but turns out I did #include print.c :sob To test the project I have all the instructions on the release page linked on the demo button and the download to os.bin which includes all the parts with the assembly bootloader, the kernel and the filesystem’s .img file. You also require qemu to be installed and if your on windows PLEASE just use wsl it’s a pain to use qemu on windows natively.", "repo": "https://github.com/jphwba/jbootloader", "demo": "https://github.com/jphwba/jbootloader/releases/tag/v1.1"}]...
|
[{"post_id": 48842, "ts": "202 [{"post_id": 48842, "ts": "2026-08-17T12:01:13Z", "hours": 7.0572, "text": "So what I implemented was a keyboard event queue, a shell, IDE driver for a FAT-16 readonly filesystem and a preemptive scheduler. I didn’t add a true ring3 userspace in this because I need GDT ring transitions and a syscall ABI that is annoying and I don’t have the legal brain capacity in my skillset to do it. This is now capable of switching into 32-bit protected mode and directly loads the C kernel into memory. I also made cleanups to kernel.c/h and fixed some minor build errors."}, {"post_id": 21713, "ts": "2026-07-03T20:08:33Z", "hours": 4.1536, "text": "Kernel Refactor One of the biggest changes in this update was reorganising the kernel into individual components rather than keeping everything inside a single source file.The project now contains dedicated modules for: VGA text output Interrupt Descriptor Table (IDT) Interrupt Service Routines (ISR) Hardware Interrupt Requests (IRQ) Programmable Interrupt Controller (PIC) Programmable Interval Timer (PIT) Keyboard driver Low-level I/O helpers Interrupt Descriptor Table The kernel now builds and installs a complete Interrupt Descriptor Table (IDT) containing all 256 entries. CPU exceptions (ISR 0-31) are fully registered, while IRQs from the Programmable Interrupt Controller occupy vectors 32-47 after remapping. Once the IDT has been loaded using lidt, the kernel is capable of safely responding to both processor exceptions and hardware interrupts. CPU Exception Handling Proper exception handling has now been implemented. Instead of the CPU immediately crashing with no feedback, exceptions are caught and displayed on screen with useful debugging information including: Exception name Interrupt number Error codeInstruction pointer (EIP) Code segment CPU flags After reporting the error the kernel safely halts, making debugging much easier during development. Programmable Interrupt Controller (PIC) The legacy PIC is now fully initialised and remapped away from the CPU exception vectors .Support has also been added for: Sending End Of Interrupt (EOI) commands Masking individual IRQ lines Unmasking IRQs when drivers are ready Programmable Interval Timer (PIT) The PIT has now been configured as the kernel’s first hardware timer. It generates periodic interrupts at a configurable frequency and maintains a continuously increasing tick counter. While simple for now, this timer will eventually become the basis for delays Keyboard Driver Perhaps the most noticeable improvement in this update is the addition of the first real hardware driver. The keyboard driver: Receives hardware interrupts on IRQ1 Reads raw PS/2 scancodes Translates scancodes into ASCII characters Buffers incoming keystrokes Allows the kernel to read characters without directly accessing hardware every time Improved VGA Terminal The VGA text driver has also been rewritten into its own module. I added Cursor updates, Automatic scrolling, Backspace support, Colour control, Decimal output and Hexadecimal output"}, {"post_id": 19946, "ts": "2026-07-01T12:11:26Z", "hours": 1.2294, "text": "I rewrote the bootloader into two stages. Each item in stage 2 is Disk loader: disk_read_lba in stage2 is fully reusable (LBA, count, drive, destination segment:offset via a Disk Address Packet), checks for INT 13h extensions first, and retries 3 times with a disk reset between attempts, same pattern in stage1’s simpler loader. Print routine: print_string prints status lines at each stage (“loading kernel…”, “enabling A20…”, etc) plus explicit disk error and unsupported extensions messages. A20: now enabled in real mode before the protected mode switch, using the BIOS function and the fast A20 gate together. Boot drive: saved from DL at both stage1 and stage2 entry rather than hardcoded. Magic numbers: sector counts, load addresses, and buffer addresses all moved into src/config.inc, shared by both stages. E820 memory map: collected in real mode into a fixed buffer before switching to protected mode. Boot info struct: built in 32-bit mode and handed to the kernel via EBX. Matching BootInfo struct is in kernel.h. Load/execute address mismatch: The kernel now loads to physical 0x10000 in real mode, then once in flat 32-bit protected mode, stage2 copies it to 0x100000 with rep movsd before jumping there, so the linker script’s link address and the actual jump target now agree."}, {"post_id": 19889, "ts": "2026-07-01T09:57:18Z", "hours": 1.8975, "text": "Got the bootloader to load kernel.c which is empty but you can see in the gdb log in the terminal that it does load void kernel_main. Next I am going to rewrite the bootloader into a two stage bootloader so it can have a print routine, retrying disk reads, A20 handling and an E820 memory map. Since all of this isnt possible in just 510 bytes (last two bytes reserved for the boot sector signature being 0x55 and 0xAA) I will split it into two stages."}, {"post_id": 17432, "ts": "2026-06-28T06:35:23Z", "hours": 0.3608, "text": "Created a basic assembly project that prints hello world to the terminal and tested it by booting the built .bin file in qemu."}]...
|
Title: JBootloader by @jphwb | Stardance
Devlogs: Title: JBootloader by @jphwb | Stardance
Devlogs:
Created a basic assembly project that prints hello world to the terminal and tested it by booting the built .bin file in qemu.
Got the bootloader to load kernel.c which is empty but you can see in the gdb log in the terminal that it does load void kernel_main. Next I am going to rewrite the bootloader into a two stage bootloader so it can have a print routine, retrying disk reads, A20 handling and an E820 memory map. Since all of this isnt possible in just 510 bytes (last two bytes reserved for the boot sector signature being 0x55 and 0xAA) I will split it into two stages.
I rewrote the bootloader into two stages. Each item in stage 2 is Disk loader: disk_read_lba in stage2 is fully reusable (LBA, count, drive, destination segment:offset via a Disk Address Packet), checks for INT 13h extensions first, and retries 3 times with a disk reset between attempts, same pattern in stage1’s simpler loader. Print routine: print_string prints status lines at each stage (“loading kernel…”, “enabling A20…”, etc) plus explicit disk error and unsupported extensions messages. A20: now enabled in real mode before the protected mode switch, using the BIOS function and the fast A20 gate together. Boot drive: saved from DL at both stage1 and stage2 entry rather than hardcoded. Magic numbers: sector counts, load addresses, and buffer addresses all moved into src/config.inc, shared by both stages. E820 memory map: collected in real mode into a fixed buffer before switching to protected mode. Boot info struct: built in 32-bit mode and handed to the kernel via EBX. Matching BootInfo struct is in kernel.h. Load/execute address mismatch: The kernel now loads to physical 0x10000 in real mode, then once in flat 32-bit protected mode, stage2 copies it to 0x100000 with rep movsd before jumping there, so the linker script’s link address and the actual jump target now agree.
Kernel Refactor One of the biggest changes in this update was reorganising the kernel into individual components rather than keeping everything inside a single source file.The project now contains dedicated modules for: VGA text output Interrupt Descriptor Table (IDT) Interrupt Service Routines (ISR) Hardware Interrupt Requests (IRQ) Programmable Interrupt Controller (PIC) Programmable Interval Timer (PIT) Keyboard driver Low-level I/O helpers Interrupt Descriptor Table The kernel now builds and installs a complete Interrupt Descriptor Table (IDT) containing all 256 entries. CPU exceptions (ISR 0-31) are fully registered, while IRQs from the Programmable Interrupt Controller occupy vectors 32-47 after remapping. Once the IDT has been loaded using lidt, the kernel is capable of safely responding to both processor exceptions and hardware interrupts. CPU Exception Handling Proper exception handling has now been implemented. Instead of the CPU immediately crashing with no feedback, exceptions are caught and displayed on screen with useful debugging information including: Exception name Interrupt number Error codeInstruction pointer (EIP) Code segment CPU flags After reporting the error the kernel safely halts, making debugging much easier during development. Programmable Interrupt Controller (PIC) The legacy PIC is now fully initialised and remapped away from the CPU exception vectors .Support has also been added for: Sending End Of Interrupt (EOI) commands Masking individual IRQ lines Unmasking IRQs when drivers are ready Programmable Interval Timer (PIT) The PIT has now been configured as the kernel’s first hardware timer. It generates periodic interrupts at a configurable frequency and maintains a continuously increasing tick counter. While simple for now, this timer will eventually become the basis for delays Keyboard Driver Perhaps the most noticeable improvement in this update is the addition of the first real hardware driver. The keyboard driver: Receives hardware interrupts on IRQ1 Reads raw PS/2 scancodes Translates scancodes into ASCII characters Buffers incoming keystrokes Allows the kernel to read characters without directly accessing hardware every time Improved VGA Terminal The VGA text driver has also been rewritten into its own module. I added Cursor updates, Automatic scrolling, Backspace support, Colour control, Decimal output and Hexadecimal output
So what I implemented was a keyboard event queue, a shell, IDE driver for a FAT-16 readonly filesystem and a preemptive scheduler. I didn’t add a true ring3 userspace in this because I need GDT ring transitions and a syscall ABI that is annoying and I don’t have the legal brain capacity in my skillset to do it. This is now capable of switching into 32-bit protected mode and directly loads the C kernel into memory. I also made cleanups to kernel.c/h and fixed some minor build errors.
Ships:
JBootloader is a custom bootloader and kernel with a shell that has basic functionality made in c and assembly x86_64. The kernel currently can load a fat-16 filesystem image in read only mode and commands such as cat and ls can be used. Everything that is required to boot is all compiled into os.bin (Releases) including the kernel, stage1/stage2.asm and the fat16 filesystem which is designed to work on QEMU. This project isnt made to be actually used (obviously) but its more of a learning journey helping me learn how a bootloader and kernel work. Initially I wanted to just make a bootloader like GRUB that could boot any operating system but at the time I was overestimating my skills in asm so I decided to steer off and make a simple kernel instead. The challenging part was the agonizing “why won’t the fat-16 partition show up” because it just would show up no matter what I did so I pulled up gemini and put in my fat_16.c code and asked it to edit my code to push out any debugging info to the vm so that I’d be able to see what went wrong, using this I overcame my problems. Also at the start I spent almost an hour trying to find an issue since my C/C++ extension in VS wasn’t working and my build command just refused to tell me what was wrong but turns out I did #include print.c :sob To test the project I have all the instructions on the release page linked on the demo button and the download to os.bin which includes all the parts with the assembly bootloader, the kernel and the filesystem’s .img file. You also require qemu to be installed and if your on windows PLEASE just use wsl it’s a pain to use qemu on windows natively....
|
1793537782.578875
|
1793624182.578875
|
|
46583
|
Fol by @vires | Stardance
|
vires
|
1
|
9.41
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T05:16:14Z", "hours": 7.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "FOL 🤖 FOL is my personal AI assistant for Mac. I wanted to make something like JARVIS from Iron Man, but actually useful in real life. FOL lives in the notch of my MacBook and can help me control my computer, open apps and projects, search the web, work with emails and my calendar, remember things, and understand what I’m doing. For example, I can say: “Open Safari and find the latest OpenAI news.” FOL can open Safari, search for the information, and give me a summary. Or I can say: “Remember that I need to send the report tomorrow.” And FOL saves it to my memory. I built FOL because I don’t want AI to be just a chat box where you type questions. I want it to actually do things for me . Right now it has voice, memory, screen context, desktop control, web search, Gmail, Google Calendar, and an agent system that lets it choose the right tools by itself. It has 989+ tests and several complete real-life scenarios already working. This is still the beginning. My goal is to make FOL feel like a real personal AI living on my Mac. See → Understand → Think → Act → Remember. 🚀", "repo": "https://github.com/abdulakimabdimanapov-rgb/FOL", "demo": "https://github.com/abdulakimabdimanapov-rgb/FOL/releases/tag/v1.3.0"}]...
|
[{"post_id": 52978, "ts": "202 [{"post_id": 52978, "ts": "2026-08-23T11:02:09Z", "hours": 2.4147, "text": "I do installer for macOS, Windows and Linux"}, {"post_id": 50783, "ts": "2026-08-20T05:08:54Z", "hours": 2.0944, "text": "Hi everyone it is Fol"}, {"post_id": 49387, "ts": "2026-08-18T05:36:34Z", "hours": 2.1972, "text": "Hi everyone this is my own anolog Jarvis Asistent (sorry my grammar )"}, {"post_id": 47975, "ts": "2026-08-16T07:53:14Z", "hours": 2.3142, "text": "Fol"}, {"post_id": 45322, "ts": "2026-08-11T18:15:53Z", "hours": 0.6169, "text": "🤖 FOL — My Personal AI Assistant for macOSI’m building FOL, a personal AI assistant for macOS inspired by the idea of a real JARVIS.FOL is designed to live directly on my MacBook through a native Notch UI and interact with my computer, applications, browser, files, memory, and tools.🚀 What I’m buildingFOL can already:🖥️ Control the desktop and interact with applications🌐 Control Chrome and Safari👁️ Understand what is happening on the screen🎙️ Support voice interaction🧠 Store and retrieve long-term memory🗂️ Integrate with Obsidian as a personal knowledge base🤖 Use an agent loop to select and execute tools✉️ Work with productivity tools such as Gmail and Google Calendar⚡ Stream responses through SSE🔐 Require confirmation before risky actions🧩 Use a centralized Tool Registry instead of scattered tool definitions🔄 Automatically fall back between LLM providers when one fails🏠 Support local models through Ollama🏗️ ArchitectureThe project is built around several services:SwiftUI → Orchestrator → LLM Router → Tool Registry → Agent Server / Memory / Productivity / FOL CoreThe current architecture uses:Python + FastAPILiteLLMSwiftUIReact / Next.jsPyAutoGUIChrome CDPOllamaObsidianSSE / WebSocket🧠 Recent progressI’ve been rebuilding the architecture phase by phase.Phase 3: unified LLM routing and deterministic fallback.Phase 4: centralized ToolRegistry + confirmation and safety gates.Phase 5: migrated all 50 tools to registry-attached handlers.Phase 6: unified memory and remaining LLM consumers around the canonical architecture.The project currently has 1000+ automated Python tests, live end-to-end scenarios, and security regression tests.🔐 Safety is a major part of FOLFOL isn’t supposed to blindly execute everything an LLM asks for.High-risk actions such as clicking, typing, sending messages, shell commands, Safari JavaScript, and cookie operations can require explicit user confirmation.The confirmation system is enforced by code, not by the model itself.Unknown tools fail closed, approvals are bound to the exact tool-call arguments, and secrets are scrubbed before being written to memory.🎯 What I’m working on nowI’m continuing the architecture migration and working toward a more unified FOL core where:Memory + LLM + Tools + Agents + UI + Safetywork together as one coherent system.The long-term goal is simple:Build a personal AI that doesn’t just chat with me, but can actually understand my environment, remember context, use tools, and help me operate my computer.This is still a work in progress, but FOL is becoming much more than a chatbot. 🚀#Stardance #HackClub #AI #macOS #Python #SwiftUI #LLM #AIAssistant #OpenSource"}]...
|
Title: Fol by @vires | Stardance
Devlogs:
🤖 FOL — Title: Fol by @vires | Stardance
Devlogs:
🤖 FOL — My Personal AI Assistant for macOSI’m building FOL, a personal AI assistant for macOS inspired by the idea of a real JARVIS.FOL is designed to live directly on my MacBook through a native Notch UI and interact with my computer, applications, browser, files, memory, and tools.🚀 What I’m buildingFOL can already:🖥️ Control the desktop and interact with applications🌐 Control Chrome and Safari👁️ Understand what is happening on the screen🎙️ Support voice interaction🧠 Store and retrieve long-term memory🗂️ Integrate with Obsidian as a personal knowledge base🤖 Use an agent loop to select and execute tools✉️ Work with productivity tools such as Gmail and Google Calendar⚡ Stream responses through SSE🔐 Require confirmation before risky actions🧩 Use a centralized Tool Registry instead of scattered tool definitions🔄 Automatically fall back between LLM providers when one fails🏠 Support local models through Ollama🏗️ ArchitectureThe project is built around several services:SwiftUI → Orchestrator → LLM Router → Tool Registry → Agent Server / Memory / Productivity / FOL CoreThe current architecture uses:Python + FastAPILiteLLMSwiftUIReact / Next.jsPyAutoGUIChrome CDPOllamaObsidianSSE / WebSocket🧠 Recent progressI’ve been rebuilding the architecture phase by phase.Phase 3: unified LLM routing and deterministic fallback.Phase 4: centralized ToolRegistry + confirmation and safety gates.Phase 5: migrated all 50 tools to registry-attached handlers.Phase 6: unified memory and remaining LLM consumers around the canonical architecture.The project currently has 1000+ automated Python tests, live end-to-end scenarios, and security regression tests.🔐 Safety is a major part of FOLFOL isn’t supposed to blindly execute everything an LLM asks for.High-risk actions such as clicking, typing, sending messages, shell commands, Safari JavaScript, and cookie operations can require explicit user confirmation.The confirmation system is enforced by code, not by the model itself.Unknown tools fail closed, approvals are bound to the exact tool-call arguments, and secrets are scrubbed before being written to memory.🎯 What I’m working on nowI’m continuing the architecture migration and working toward a more unified FOL core where:Memory + LLM + Tools + Agents + UI + Safetywork together as one coherent system.The long-term goal is simple:Build a personal AI that doesn’t just chat with me, but can actually understand my environment, remember context, use tools, and help me operate my computer.This is still a work in progress, but FOL is becoming much more than a chatbot. 🚀#Stardance #HackClub #AI #macOS #Python #SwiftUI #LLM #AIAssistant #OpenSource
Fol
Hi everyone this is my own anolog Jarvis Asistent (sorry my grammar )
Hi everyone it is Fol
I do installer for macOS, Windows and Linux
Ships:
FOL 🤖 FOL is my personal AI assistant for Mac. I wanted to make something like JARVIS from Iron Man, but actually useful in real life. FOL lives in the notch of my MacBook and can help me control my computer, open apps and projects, search the web, work with emails and my calendar, remember things, and understand what I’m doing. For example, I can say: “Open Safari and find the latest OpenAI news.” FOL can open Safari, search for the information, and give me a summary. Or I can say: “Remember that I need to send the report tomorrow.” And FOL saves it to my memory. I built FOL because I don’t want AI to be just a chat box where you type questions. I want it to actually do things for me . Right now it has voice, memory, screen context, desktop control, web search, Gmail, Google Calendar, and an agent system that lets it choose the right tools by itself. It has 989+ tests and several complete real-life scenarios already working. This is still the beginning. My goal is to make FOL feel like a real personal AI living on my Mac. See → Understand → Think → Act → Remember. 🚀...
|
1792878490.3371441
|
1793742490.3371441
|
|
50464
|
WebOS EVAN by @anura | Stardance
|
anura
|
2
|
10.0
|
36
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T05:09:17Z", "hours": 5.0, "multiplier": 6.95, "payout": 36, "blessing": "neutral", "text": "Upgraded my nerv terminal for WebOS 2 added lilith wallpaper, sound effect, pilot log notes that save, classified file system with lore, min/max window control, custom right click menu, aeva launch sequence from terminal, and a magi override vote when you clear the angel alert… the typoss almost killed mee tying to fix itt(ToT)…", "repo": "https://github.com/anuragtamang095-coder/Webos", "demo": "https://anuragtamang095-coder.github.io/Webos/"}, {"ship_num": 2, "ts": "2026-08-18T03:51:01Z", "hours": 5.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I made my very firsttt OS (TOT).. very proud of my self… did use ai when i waas stuck in debugging.. this was very gud experience and i look forward on the WebOS 2.", "repo": "https://github.com/anuragtamang095-coder/Webos", "demo": "https://anuragtamang095-coder.github.io/Webos/"}]...
|
[{"post_id": 50782, "ts": "202 [{"post_id": 50782, "ts": "2026-08-20T05:05:01Z", "hours": 0.6608, "text": "Finallyyy done withhh ittt(T-T)..this project GOT ME CRYINGG"}, {"post_id": 50768, "ts": "2026-08-20T04:15:35Z", "hours": 0.5917, "text": "Just came back from Tuition and checked my WebOS and found that The terminal commands and there work was so bland so i am trying to add some spice.. lets see how it turns out.."}, {"post_id": 50711, "ts": "2026-08-20T01:23:29Z", "hours": 0.5081, "text": "last night just was not able to finish the minimize and maximize button had to wake up early in a day off.. maybe it works.. i check.."}, {"post_id": 50464, "ts": "2026-08-19T16:58:51Z", "hours": 0.4222, "text": "I think enough for today almost done with the minimize and maximize window buttonnnn.."}, {"post_id": 50435, "ts": "2026-08-19T16:30:57Z", "hours": 1.4847, "text": "Me hitting an error..meanwhile my JS: loda Notes.. LMAOO *Skull-emojii"}, {"post_id": 50327, "ts": "2026-08-19T14:26:57Z", "hours": 0.8247, "text": "Added the boot, click, angel alert sound effectt… right click added(T-T) Finally getting the hang of itt.. feeling like finishing this project in one sittinggg.."}, {"post_id": 50290, "ts": "2026-08-19T13:20:47Z", "hours": 0.7125, "text": "Long day at school and exhaustion is fixed when i see progress of my code(^ w ^).. done with the wallpaper now looks pretty good.. thinking of adding 5 new features on this WebOS as for its mission lets see how much i can.. Que sera sera…"}, {"post_id": 49333, "ts": "2026-08-18T03:00:34Z", "hours": 0.5225, "text": "So, the spaces matter..(-_-)my lazzy aahh didn’t add spaces so now i got another 1/2 hr of debugging maybe.."}, {"post_id": 49314, "ts": "2026-08-18T01:58:32Z", "hours": 0.5042, "text": "Finallyyy Donew with the os.. gonna ship the project after final checking..(TOT)"}, {"post_id": 49302, "ts": "2026-08-18T01:19:37Z", "hours": 0.3039, "text": "just woke up and Started the JS, he time consume is half hour more than i planned (-_-)… IS there anything that can help me code faster???"}, {"post_id": 49019, "ts": "2026-08-17T16:43:38Z", "hours": 3.8872, "text": "Dyaammm!! it’s difficult yet addictive..done with the HTML and CSS now JS needs to be done!! EVANGELION OS on the wayy!!"}]...
|
Title: WebOS EVAN by @anura | Stardance
Devlogs:
Title: WebOS EVAN by @anura | Stardance
Devlogs:
Dyaammm!! it’s difficult yet addictive..done with the HTML and CSS now JS needs to be done!! EVANGELION OS on the wayy!!
just woke up and Started the JS, he time consume is half hour more than i planned (-_-)… IS there anything that can help me code faster???
Finallyyy Donew with the os.. gonna ship the project after final checking..(TOT)
So, the spaces matter..(-_-)my lazzy aahh didn’t add spaces so now i got another 1/2 hr of debugging maybe..
Long day at school and exhaustion is fixed when i see progress of my code(^ w ^).. done with the wallpaper now looks pretty good.. thinking of adding 5 new features on this WebOS as for its mission lets see how much i can.. Que sera sera…
Added the boot, click, angel alert sound effectt… right click added(T-T) Finally getting the hang of itt.. feeling like finishing this project in one sittinggg..
Me hitting an error..meanwhile my JS: loda Notes.. LMAOO *Skull-emojii
I think enough for today almost done with the minimize and maximize window buttonnnn..
last night just was not able to finish the minimize and maximize button had to wake up early in a day off.. maybe it works.. i check..
Just came back from Tuition and checked my WebOS and found that The terminal commands and there work was so bland so i am trying to add some spice.. lets see how it turns out..
Finallyyy done withhh ittt(T-T)..this project GOT ME CRYINGG
Ships:
I made my very firsttt OS (TOT).. very proud of my self… did use ai when i waas stuck in debugging.. this was very gud experience and i look forward on the WebOS 2.
Upgraded my nerv terminal for WebOS 2 added lilith wallpaper, sound effect, pilot log notes that save, classified file system with lore, min/max window control, custom right click menu, aeva launch sequence from terminal, and a magi override vote when you clear the angel alert… the typoss almost killed mee tying to fix itt(ToT)…...
|
1793105308.9277956
|
1793969308.9277956
|
|
49421
|
My webside by @projectroshan917 | Stardance
|
projectroshan917
|
1
|
2.0
|
10
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T03:39:37Z", "hours": 2.0, "multiplier": 1.01, "payout": 10, "blessing": "neutral", "text": "I made an personal portfolio webside where my info is present.", "repo": "https://github.com/Roshan-coder123/My-website-", "demo": "https://roshan-coder123.github.io/My-website-/"}]...
|
[{"post_id": 50757, "ts": "202 [{"post_id": 50757, "ts": "2026-08-20T03:32:55Z", "hours": 1.7203, "text": "My person website"}]...
|
Title: My webside by @projectroshan917 | Stardance Title: My webside by @projectroshan917 | Stardance
Devlogs:
My person website
Ships:
I made an personal portfolio webside where my info is present....
|
1792878696.9654155
|
1793742696.9654155
|
|
29625
|
Investie by @knightfallShashy | Stardance
|
knightfallShashy
|
1
|
25.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T03:32:51Z", "hours": 25.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Investie is a full stack investment based research app. You can search any stock, and get an AI summary of its financials and a real DCF valuation with proper market data. I built the stack myself with reflex as the frontend, fastAPI backend, and the postures via sup abase. I used finance for the data, and the anthropic API for the summaries. The hardest part was getting it deployed as it took a really long time to do that especially with the python version mismatch. My proudest accomplishment is probably the DCF engine and the AI summaries page.", "repo": "https://github.com/Shashyster/Investie", "demo": "https://frontend-silver-sun.reflex.run"}]...
|
[{"post_id": 50755, "ts": "202 [{"post_id": 50755, "ts": "2026-08-20T03:29:04Z", "hours": 1.6825, "text": "This is probably my last devlog for Investie. I’ve finished deploying the reflex frontend and fastAPI backend to Render and Reflex. I spent a frustratingly long time fixing the deploy blockers from Yfinance’s rate limiting. I also spent a good amount of time writing an entire README for the whole project. Requirements.txt was a disaster and I spent a long time fixing it and generating it again. Lastly, I fixed the frontend properties and gave it a proper error message for when yahoo limits requests."}, {"post_id": 50122, "ts": "2026-08-19T06:27:21Z", "hours": 2.4925, "text": "First of all I fixed the market page, instead of everything crammed to the left, it’s in the middle now and looks a lot more organized. Secondly I finally connected Anthropic’s API key to the project in- order to give stock summaries. The prompt for summaries took way too long because I kept changing it. The work done really put into perspective how long it takes to come up with the right prompt so that your tokens aren’t being burned. Im planning on investing a lot more money for the API tokens. lastly, Im starting to work the stock cards and making them clickable and a few other changes here and there."}, {"post_id": 47321, "ts": "2026-08-15T06:19:58Z", "hours": 3.2797, "text": "To start off I fixed the number formatting on my DCF, which was initially showing me just raw numbers. I also found out that apparent reflex state vars doesn’t support f- string format lines which was a bit annoying. The next big thing is that I built a full company name feature. Instead of just searching the ticker you can search the company name and a list will pop up. That specifically took a long time to build. Lastly I wired real market data to the stock cards on the homepage, giving it a realism effect. I still have plans on taking it further and making them clickable."}, {"post_id": 38012, "ts": "2026-07-31T04:31:40Z", "hours": 3.515, "text": "I built a pipeline for the DCF valuation logic to connect through the FastAPI endpoints (aka Get/fmp/ticker/get_full_dcf_valuation). I also fixed an issue on the startup with my backend server as fastapi was downgraded while starlette was upgraded. Additionally, I’ve finally made up my mind instead of using fmp by paying $19 a month, im going with yfinance’s free library. The switch made me change ALOT of Fmp code around. Lastly, I’ve started wiring the reflex frontend with the backend and wrote my “state” class. Overall I built new endpoints, changed previous ones, and worked on existing ones with major bug fixes in the middle."}, {"post_id": 33867, "ts": "2026-07-24T04:11:49Z", "hours": 3.6742, "text": "Big frontend push this session. I finished building out all five core pages: Home, Markets, Watchlist, About, and Contact. I also wired up a shared navbar with real routing. The ticker research input now lives on the Markets page instead of Home, freeing up the homepage for additional details. Built out the Home page hero to match a Bloomberg Terminal-inspired design: a dark section with a faint SVG chart-line pattern behind the text, and an “Explore Markets” button that actually routes to /markets. Also I made a scrollable charts section In the top. Overall it was filled with a lot of complicated Reflex design and painful bugs. I’m going to connect the backend and frontend soon though."}, {"post_id": 32772, "ts": "2026-07-22T04:37:30Z", "hours": 2.7183, "text": "Wrapped up the last piece of the core DCF math, then started the actual website since real cash flow data is still held behind the FMP Starter upgrade. I built the WACC math which ties into the Dcf calculation. Additionally, I have started my frontend by installing reflex and starting to work on the site itself. It’s still in it’s starting stages but took a lot of bug fixing."}, {"post_id": 30535, "ts": "2026-07-18T06:01:16Z", "hours": 1.83, "text": "Picked up where I left off with two small FMP cleanup items, then spent most of this session building the actual valuation math. Found out the $19/mo Starter tier does not extend that 5-year cap, it only unlocks the balance sheet/cash flow endpoints and more bandwidth. Every function got hand verified against math I worked out on paper before trusting the code, there is nothing that went in without me being able to explain why the numbers came out the way they did. I have much more screenshots, but didn’t have the time to paste them."}, {"post_id": 26837, "ts": "2026-07-11T21:09:14Z", "hours": 1.8503, "text": "I built out functions to pull income statements, balance sheets, and cash flow statements straight from FMP, following the same pattern as my earlier company profile integration. Income statement worked immediately with real multi-year data for Apple. Balance sheet and cash flow both came back with a 404 though. Apparently, my free-tier FMP account likely doesn’t include full financial statement access, based on comparing their pricing page (that data seems to be behind their $19/mo Starter plan). Rather than paying right away, I’m holding off and treating this as a “figure out before launch” decision."}, {"post_id": 22672, "ts": "2026-07-05T07:05:31Z", "hours": 1.9417, "text": "I finally built out my database session and my first company table model. Aditionally I also wrote a POST/ companies endpoint that takes data, validates it, and saves it to the database. Lastly after so much fixing I finally got a clean 200 response back with Microsoft’s data and a real database ID, so the full loop — request → validate → save → confirm — is working end to end now."}, {"post_id": 20534, "ts": "2026-07-02T04:41:36Z", "hours": 1.7736, "text": "This is my first devlog for Investie. So far I finished installing all my requirements, I made a Fast API server with a few words on it. Additionally, I ended up getting a bunch of important files made that are needed for the project. Lastly, I signed up with SupaBase to launch my site through their network and use it for API key connections. The main part of today was debugging, which took a painfully long time."}]...
|
Title: Investie by @knightfallShashy | Stardance
Title: Investie by @knightfallShashy | Stardance
Devlogs:
This is my first devlog for Investie. So far I finished installing all my requirements, I made a Fast API server with a few words on it. Additionally, I ended up getting a bunch of important files made that are needed for the project. Lastly, I signed up with SupaBase to launch my site through their network and use it for API key connections. The main part of today was debugging, which took a painfully long time.
I finally built out my database session and my first company table model. Aditionally I also wrote a POST/ companies endpoint that takes data, validates it, and saves it to the database. Lastly after so much fixing I finally got a clean 200 response back with Microsoft’s data and a real database ID, so the full loop — request → validate → save → confirm — is working end to end now.
I built out functions to pull income statements, balance sheets, and cash flow statements straight from FMP, following the same pattern as my earlier company profile integration. Income statement worked immediately with real multi-year data for Apple. Balance sheet and cash flow both came back with a 404 though. Apparently, my free-tier FMP account likely doesn’t include full financial statement access, based on comparing their pricing page (that data seems to be behind their $19/mo Starter plan). Rather than paying right away, I’m holding off and treating this as a “figure out before launch” decision.
Picked up where I left off with two small FMP cleanup items, then spent most of this session building the actual valuation math. Found out the $19/mo Starter tier does not extend that 5-year cap, it only unlocks the balance sheet/cash flow endpoints and more bandwidth. Every function got hand verified against math I worked out on paper before trusting the code, there is nothing that went in without me being able to explain why the numbers came out the way they did. I have much more screenshots, but didn’t have the time to paste them.
Wrapped up the last piece of the core DCF math, then started the actual website since real cash flow data is still held behind the FMP Starter upgrade. I built the WACC math which ties into the Dcf calculation. Additionally, I have started my frontend by installing reflex and starting to work on the site itself. It’s still in it’s starting stages but took a lot of bug fixing.
Big frontend push this session. I finished building out all five core pages: Home, Markets, Watchlist, About, and Contact. I also wired up a shared navbar with real routing. The ticker research input now lives on the Markets page instead of Home, freeing up the homepage for additional details. Built out the Home page hero to match a Bloomberg Terminal-inspired design: a dark section with a faint SVG chart-line pattern behind the text, and an “Explore Markets” button that actually routes to /markets. Also I made a scrollable charts section In the top. Overall it was filled with a lot of complicated Reflex design and painful bugs. I’m going to connect the backend and frontend soon though.
I built a pipeline for the DCF valuation logic to connect through the FastAPI endpoints (aka Get/fmp/ticker/get_full_dcf_valuation). I also fixed an issue on the startup with my backend server as fastapi was downgraded while starlette was upgraded. Additionally, I’ve finally made up my mind instead of using fmp by paying $19 a month, im going with yfinance’s free library. The switch made me change ALOT of Fmp code around. Lastly, I’ve started wiring the reflex frontend with the backend and wrote my “state” class. Overall I built new endpoints, changed previous ones, and worked on existing ones with major bug fixes in the middle.
To start off I fixed the number formatting on my DCF, which was initially showing me just raw numbers. I also found out that apparent reflex state vars doesn’t support f- string format lines which was a bit annoying. The next big thing is that I built a full company name feature. Instead of just searching the ticker you can search the company name and a list will pop up. That specifically took a long time to build. Lastly I wired real market data to the stock cards on the homepage, giving it a realism effect. I still have plans on taking it further and making them clickable.
First of all I fixed the market page, instead of everything crammed to the left, it’s in the middle now and looks a lot more organized. Secondly I finally connected Anthropic’s API key to the project in- order to give stock summaries. The prompt for summaries took way too long because I kept changing it. The work done really put into perspective how long it takes to come up with the right prompt so that your tokens aren’t being burned. Im planning on investing a lot more money for the API tokens. lastly, Im starting to work the stock cards and making them clickable and a few other changes here and there.
This is probably my last devlog for Investie. I’ve finished deploying the reflex frontend and fastAPI backend to Render and Reflex. I spent a frustratingly long time fixing the deploy blockers from Yfinance’s rate limiting. I also spent a good amount of time writing an entire README for the whole project. Requirements.txt was a disaster and I spent a long time fixing it and generating it again. Lastly, I fixed the frontend properties and gave it a proper error message for when yahoo limits requests.
Ships:
Investie is a full stack investment based research app. You can search any stock, and get an AI summary of its financials and a real DCF valuation with proper market data. I built the stack myself with reflex as the frontend, fastAPI backend, and the postures via sup abase. I used finance for the data, and the anthropic API for the summaries. The hardest part was getting it deployed as it took a really long time to do that especially with the python version mismatch. My proudest accomplishment is probably the DCF engine and the AI summaries page....
|
1793505167.2728488
|
1793591567.2728488
|
|
51703
|
Textwise by @Mov | Stardance
|
Mov
|
1
|
7.59
|
39
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T02:16:15Z", "hours": 7.0, "multiplier": 5.67, "payout": 39, "blessing": "neutral", "text": "Released v1.0.0, Includes: image generation from images and procedural gradients custom bitmap or TTF fonts 6 fonts bundled in the application procedural palettes of any size", "repo": "https://github.com/man-o-valor/textwise", "demo": "https://github.com/man-o-valor/textwise/releases/"}]...
|
[{"post_id": 51213, "ts": "202 [{"post_id": 51213, "ts": "2026-08-20T18:56:16Z", "hours": 0.5933, "text": "Added automatic palette sizing powered by the CIELAB color space! Now when not specifying a palette size it tries to find a good size to use automatically. Because of the color conversion, it’s quite a bit slower. I’ll look into optimizations soon Also, I added Kitchen Sink, another fun 6x8 pixel font. Below is an image auto-assigned to only need 4 COLORS– and it looks super!"}, {"post_id": 50725, "ts": "2026-08-20T02:08:26Z", "hours": 6.9483, "text": "Released v1.0 of Textwise! Textwise is a TUI-based program powered by Node that generates “textmode”-styled mosaics based on provided images or procedurally generated gradients. Allows processing existing images and making procedural gradients Comes bundled with 6 fonts Palettes are procedural and size can be specified Supports solid and per-glyph background colors"}]...
|
Title: Textwise by @Mov | Stardance
Devlogs:
Rele Title: Textwise by @Mov | Stardance
Devlogs:
Released v1.0 of Textwise! Textwise is a TUI-based program powered by Node that generates “textmode”-styled mosaics based on provided images or procedurally generated gradients. Allows processing existing images and making procedural gradients Comes bundled with 6 fonts Palettes are procedural and size can be specified Supports solid and per-glyph background colors
Added automatic palette sizing powered by the CIELAB color space! Now when not specifying a palette size it tries to find a good size to use automatically. Because of the color conversion, it’s quite a bit slower. I’ll look into optimizations soon Also, I added Kitchen Sink, another fun 6x8 pixel font. Below is an image auto-assigned to only need 4 COLORS– and it looks super!
Ships:
Released v1.0.0, Includes: image generation from images and procedural gradients custom bitmap or TTF fonts 6 fonts bundled in the application procedural palettes of any size...
|
1793526658.4571774
|
1793613058.4571774
|
|
51610
|
Math Solver by @ameeksinghv | Stardance
|
ameeksinghv
|
1
|
2.31
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-20T00:35:33Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Its a calculator! U can calculate. idrk i just made this cuz i had a really simple one laying around and i decided why not.", "repo": "https://github.com/singhameek/math", "demo": "https://github.com/singhameek/math/releases/tag/v1"}]...
|
[{"post_id": 51827, "ts": "202 [{"post_id": 51827, "ts": "2026-08-21T17:59:10Z", "hours": 0.2781, "text": "Finished sim linear equations, added quadratics!"}, {"post_id": 51376, "ts": "2026-08-21T01:13:26Z", "hours": 0.4958, "text": "I’ve decided not to make the calculator a site yet, because i realize that this was a bit beyond my scope and that i would rather write bare python code."}, {"post_id": 51236, "ts": "2026-08-20T19:30:20Z", "hours": 0.5311, "text": "Moved the app from Streamlit to Reflex"}, {"post_id": 50688, "ts": "2026-08-20T00:23:22Z", "hours": 0.3733, "text": "I think?? im done the code but idk how to fix this imma research that"}, {"post_id": 50595, "ts": "2026-08-19T20:13:21Z", "hours": 0.9897, "text": "idk why i think cuz i havent used some of the dependencies but that shouldnt do that so we’ll figure it out guys"}]...
|
Title: Math Solver by @ameeksinghv | Stardance
De Title: Math Solver by @ameeksinghv | Stardance
Devlogs:
idk why i think cuz i havent used some of the dependencies but that shouldnt do that so we’ll figure it out guys
I think?? im done the code but idk how to fix this imma research that
Moved the app from Streamlit to Reflex
I’ve decided not to make the calculator a site yet, because i realize that this was a bit beyond my scope and that i would rather write bare python code.
Finished sim linear equations, added quadratics!
Ships:
Its a calculator! U can calculate. idrk i just made this cuz i had a really simple one laying around and i decided why not....
|
1793515897.7934983
|
1793602297.7934983
|
|
51647
|
Personal Site + Backend by @MaddiefromBelow | Star Personal Site + Backend by @MaddiefromBelow | Stardance...
|
MaddiefromBelow
|
1
|
7.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T23:25:46Z", "hours": 7.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Made a personal website, and more crucially its backend (in rust btw). Particularly fun was navigating clippy on the strictest lint settings for the first time, which had me discarding a lot of bad practices i’ve implemented before, and (maybe) learning new bad habits to replace them :3 Really liked hacking together a system to fulfill specifically my own purposes, and the delightful refrain of “out of scope” every time something threatened to lure me into the Torment Hole. Maybe i’ll work on this more at a later date but it’s fulfilled all the requirements i started with so im happy to ship", "repo": "https://github.com/Madelyn-of-Hell/maddie-backend", "demo": "https://oneluckymushroom.dev"}]...
|
[{"post_id": 50642, "ts": "202 [{"post_id": 50642, "ts": "2026-08-19T22:03:07Z", "hours": 6.9417, "text": "Devlog #1: uuh…. done, I guess? OK so full disclosure: this wasn’t initially going to be a stardance project. To be perfectly honest, I’d forgotten hackclub existed until I got their weekly email, and then I realised I could retroactively submit this here. So, here it is: a custom rust server that supports not just the custom & kinda funky website I built, but also runs a multiplayer server for my as-yet unfinished game Particles, a boardgame I built based on a (no offense) kinda crummy mobile app, a writeup from the two thousands for a now inaccessible windows game, and a footnote of a game in one of those ad-riddled ‘unblocked games’ websites that my sister showed me. I may go into more depth about that game as another project, but for now I’ll only talk about the server. It’s pretty simple, to be honest: since I made it for me and my close friends, I felt no need whatsoever to try and verify the integrity of the game. There’s no anticheat, no simulation of results: all it does is handle lobbies (plaintext json files with a uuid for a name) and players (single-use UUIDs assigned on joining a lobby). All exploration of state, validation of moves, and win checking is handled on the client side. If one of my friends exploits this system, it’s going to be very funny, and I can’t wait to see how they do. Until then, everything else can happen on the client side. All the server needs to do is manage https requests. The static site is, amusingly a little more complex. Inspired by friends like circulars, carmine cassie, and ravarcheon whose websites are relatively simple and yet, it seems to me, contain a piece of their souls that feels unique and delightful, the core purpose of this site is to incite a single moment of delight. Or amusement. I’d take a raised eyebrow, to be honest. Look, I’m not a frontend developer. I kinda suck at visual design, and my website isn’t going to be labeled ‘pretty’ or ‘elegant’. What I am is creative, and a bit of a sneak. I’m someone who likes puzzles. I’m someone who likes surprising my friends. And I’m someone who’s willing to offer an objectively worse experience to the masses in order to create a cool experience for the people who found it. I’m not going to go into depth about what my ‘trick’ is here. It’s hardly a challenging riddle (in fact, I might revise it at some point to increase the challenge), but it is Neat and it provides the little jolt of excitement i’m looking for. And, crucially, while I didn’t want to use a full web framework like Actix, it required more complexity than just serving the path requested. Thus, I designed a system based around a manifest.json file. It contains a dictionary, where each key is the path requested, and each value is another dictionary containing a few pieces of information such as filepath, Mime type, and… well, if you’re curious take a peek for yourself! (Do try and figure out the trick with the website before you look at the answers, though.) anyways: i use the tiny_http crate to handle incoming requests, partially because i don’t wanna figure out certificates on my own, and partially because having requests as a struct is really handy. Then there’s the final element of the trick: the HTML part. Without saying too much, it’s two similar pages with different css and then an SVG clipping mask :3 Hope people like it !!"}]...
|
Title: Personal Site + Backend by @MaddiefromBelow Title: Personal Site + Backend by @MaddiefromBelow | Stardance
Devlogs:
Devlog #1: uuh…. done, I guess? OK so full disclosure: this wasn’t initially going to be a stardance project. To be perfectly honest, I’d forgotten hackclub existed until I got their weekly email, and then I realised I could retroactively submit this here. So, here it is: a custom rust server that supports not just the custom & kinda funky website I built, but also runs a multiplayer server for my as-yet unfinished game Particles, a boardgame I built based on a (no offense) kinda crummy mobile app, a writeup from the two thousands for a now inaccessible windows game, and a footnote of a game in one of those ad-riddled ‘unblocked games’ websites that my sister showed me. I may go into more depth about that game as another project, but for now I’ll only talk about the server. It’s pretty simple, to be honest: since I made it for me and my close friends, I felt no need whatsoever to try and verify the integrity of the game. There’s no anticheat, no simulation of results: all it does is handle lobbies (plaintext json files with a uuid for a name) and players (single-use UUIDs assigned on joining a lobby). All exploration of state, validation of moves, and win checking is handled on the client side. If one of my friends exploits this system, it’s going to be very funny, and I can’t wait to see how they do. Until then, everything else can happen on the client side. All the server needs to do is manage https requests. The static site is, amusingly a little more complex. Inspired by friends like circulars, carmine cassie, and ravarcheon whose websites are relatively simple and yet, it seems to me, contain a piece of their souls that feels unique and delightful, the core purpose of this site is to incite a single moment of delight. Or amusement. I’d take a raised eyebrow, to be honest. Look, I’m not a frontend developer. I kinda suck at visual design, and my website isn’t going to be labeled ‘pretty’ or ‘elegant’. What I am is creative, and a bit of a sneak. I’m someone who likes puzzles. I’m someone who likes surprising my friends. And I’m someone who’s willing to offer an objectively worse experience to the masses in order to create a cool experience for the people who found it. I’m not going to go into depth about what my ‘trick’ is here. It’s hardly a challenging riddle (in fact, I might revise it at some point to increase the challenge), but it is Neat and it provides the little jolt of excitement i’m looking for. And, crucially, while I didn’t want to use a full web framework like Actix, it required more complexity than just serving the path requested. Thus, I designed a system based around a manifest.json file. It contains a dictionary, where each key is the path requested, and each value is another dictionary containing a few pieces of information such as filepath, Mime type, and… well, if you’re curious take a peek for yourself! (Do try and figure out the trick with the website before you look at the answers, though.) anyways: i use the tiny_http crate to handle incoming requests, partially because i don’t wanna figure out certificates on my own, and partially because having requests as a struct is really handy. Then there’s the final element of the trick: the HTML part. Without saying too much, it’s two similar pages with different css and then an SVG clipping mask :3 Hope people like it !!
Ships:
Made a personal website, and more crucially its backend (in rust btw). Particularly fun was navigating clippy on the strictest lint settings for the first time, which had me discarding a lot of bad practices i’ve implemented before, and (maybe) learning new bad habits to replace them :3 Really liked hacking together a system to fulfill specifically my own purposes, and the delightful refrain of “out of scope” every time something threatened to lure me into the Torment Hole. Maybe i’ll work on this more at a later date but it’s fulfilled all the requirements i started with so im happy to ship...
|
1793515989.4289646
|
1793602389.4289646
|
|
46543
|
My Slack Bot by @Ditto | Stardance
|
Ditto
|
1
|
2.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T22:42:28Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I built a Slack bot that runs 24/7. I used Hack Club Nest as the server and it supports 15 commands. During development, the /dittobot-ping command stopped working, so I dug around everywhere trying to find the cause, but I couldn’t find it. It turned out that the section containing the /dittobot-ping command had been deleted during refactoring… Once I fixed that, it finally worked… I felt absolutely great…", "repo": "https://github.com/d08835149-prog/DittoBot", "demo": "https://app.slack.com/client/E09V59WQY1E/C0P5NE354?skip_today=1"}]...
|
[{"post_id": 50643, "ts": "202 [{"post_id": 50643, "ts": "2026-08-19T22:04:06Z", "hours": 2.0542, "text": "I made DittoBot because I wanted to try making my own Slack bot. I started with a simple ping command and kept adding more things like weather, calculator, jokes, random choices, and other small tools. As the code got longer, I separated each command into its own file. I also deployed the bot on Hack Club Nest, so it can stay online 24/7. I learned a lot about Slack bots, APIs, GitHub, and deploying while making it. During development, I deployed to Nest and tried running the /dittobot-ping command, but it didn’t work. I tinkered with various settings, but nothing worked properly. It turned out that the /ping command code was missing from index.js … Once I added it back, it started working correctly."}]...
|
Title: My Slack Bot by @Ditto | Stardance
Devlogs Title: My Slack Bot by @Ditto | Stardance
Devlogs:
I made DittoBot because I wanted to try making my own Slack bot. I started with a simple ping command and kept adding more things like weather, calculator, jokes, random choices, and other small tools. As the code got longer, I separated each command into its own file. I also deployed the bot on Hack Club Nest, so it can stay online 24/7. I learned a lot about Slack bots, APIs, GitHub, and deploying while making it. During development, I deployed to Nest and tried running the /dittobot-ping command, but it didn’t work. I tinkered with various settings, but nothing worked properly. It turned out that the /ping command code was missing from index.js … Once I added it back, it started working correctly.
Ships:
I built a Slack bot that runs 24/7. I used Hack Club Nest as the server and it supports 15 commands. During development, the /dittobot-ping command stopped working, so I dug around everywhere trying to find the cause, but I couldn’t find it. It turned out that the section containing the /dittobot-ping command had been deleted during refactoring… Once I fixed that, it finally worked… I felt absolutely great…...
|
1793490639.5403457
|
1793577039.5403457
|
|
46941
|
Laufen by @Vihaan15 | Stardance
|
Vihaan15
|
1
|
5.0
|
21
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T20:34:50Z", "hours": 5.0, "multiplier": 8.13, "payout": 21, "blessing": "neutral", "text": "Laufen turns your typing speed into a car race. A paragraph appears, you type it right on the screen, and your car races an AI to the finish line — correct letters move you, mistakes stop you, and a 2-minute timer keeps the pressure on. When the race ends you get your real WPM and accuracy. Built with plain HTML, CSS and vanilla JavaScript", "repo": "https://github.com/Vihaan6119/Laufen", "demo": "https://Vihaan6119.github.io/Laufen"}]...
|
[{"post_id": 50611, "ts": "202 [{"post_id": 50611, "ts": "2026-08-19T20:34:13Z", "hours": 2.2, "text": "Finished working on my project named Laufen. Why is js so annoying code?"}, {"post_id": 49831, "ts": "2026-08-18T19:39:03Z", "hours": 2.8331, "text": "V1 of my typing practice/game. Please tell me what all features I should add, things I should improve, and nice color grading also. Much appreciated if you will."}]...
|
Title: Laufen by @Vihaan15 | Stardance
Devlogs:
V Title: Laufen by @Vihaan15 | Stardance
Devlogs:
V1 of my typing practice/game. Please tell me what all features I should add, things I should improve, and nice color grading also. Much appreciated if you will.
Finished working on my project named Laufen. Why is js so annoying code?
Ships:
Laufen turns your typing speed into a car race. A paragraph appears, you type it right on the screen, and your car races an AI to the finish line — correct letters move you, mistakes stop you, and a 2-minute timer keeps the pressure on. When the race ends you get your real WPM and accuracy. Built with plain HTML, CSS and vanilla JavaScript...
|
1793325102.909035
|
1793411502.909035
|
|
51626
|
My personal site! by @potipek | Stardance
|
potipek
|
1
|
1.44
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T20:18:21Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I built my personal site, i know its not good but i am a newbie in html and css. What does my project feauture? It contains Who am I It contains a funny picture of Pavel Cimbál It has my github profile when you click on github logo", "repo": "https://github.com/tomk1440-cmyk/personal-site", "demo": "https://tomk1440-cmyk.github.io/personal-site/"}]...
|
[{"post_id": 51205, "ts": "202 [{"post_id": 51205, "ts": "2026-08-20T18:38:01Z", "hours": 0.435, "text": "Added 2 categories and a picture of my server :)"}, {"post_id": 50598, "ts": "2026-08-19T20:16:16Z", "hours": 0.7289, "text": "I built my personal website featuring my projects"}]...
|
Title: My personal site! by @potipek | Stardance
Title: My personal site! by @potipek | Stardance
Devlogs:
I built my personal website featuring my projects
Added 2 categories and a picture of my server :)
Ships:
I built my personal site, i know its not good but i am a newbie in html and css. What does my project feauture? It contains Who am I It contains a funny picture of Pavel Cimbál It has my github profile when you click on github logo...
|
1793515972.7463322
|
1793602372.7463322
|
|
51619
|
My personal website! by @raoul | Stardance
|
raoul
|
1
|
1.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T19:32:07Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I made V1 of my personal website! It has an embedded link and picture of my other project: the Michaelson Interferometer!", "repo": "https://github.com/rsourcer/Personal-Website", "demo": "https://rsourcer.github.io/Personal-Website/"}]...
|
[{"post_id": 50565, "ts": "202 [{"post_id": 50565, "ts": "2026-08-19T19:29:33Z", "hours": 0.7369, "text": "Made V1 of my (soon to be) portfolio website! I will probably iterate on it much more"}]...
|
Title: My personal website! by @raoul | Stardance
Title: My personal website! by @raoul | Stardance
Devlogs:
Made V1 of my (soon to be) portfolio website! I will probably iterate on it much more
Ships:
I made V1 of my personal website! It has an embedded link and picture of my other project: the Michaelson Interferometer!...
|
1793515951.832045
|
1793602351.832045
|
|
51615
|
Website about cats by @jenin | Stardance
|
jenin
|
1
|
1.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T19:30:15Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I shipped V1 of my site! It’s really cool and has a photo of a cat.", "repo": "https://github.com/jeninh/website-about-cats", "demo": "https://jeninh.github.io/website-about-cats/"}]...
|
[{"post_id": 50561, "ts": "202 [{"post_id": 50561, "ts": "2026-08-19T19:27:31Z", "hours": 0.6217, "text": "I made V1 of my website!!! It’s really cool and it’s about cats."}]...
|
Title: Website about cats by @jenin | Stardance
D Title: Website about cats by @jenin | Stardance
Devlogs:
I made V1 of my website!!! It’s really cool and it’s about cats.
Ships:
I shipped V1 of my site! It’s really cool and has a photo of a cat....
|
1793515910.4116204
|
1793602310.4116204
|
|
42383
|
fieldPlate by @Maarxius | Stardance
|
Maarxius
|
1
|
5.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T19:27:00Z", "hours": 5.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "fieldPlate This project started out as a joke. I had a funny thought of converting a camera tripod into an armrest. Then I thought, I have a 3D printer plate that’s flat enough. Why not turn that into a table and stick it on top of a tripod? Maybe some godforsaken photographer could use it while they’re out in the field, and thus fieldPlate was born. Unfortunately, I do not have the time to do this project the full justice it deserves. That’s definitely a full P3 thing. However, I’m pretty proud of what I’ve done. In essence, it’s a very simple project. It’s just a tripod attachment with four arms, which allows you to hold things, and it works. It’s pretty good. It also has magnetic attachments, so you don’t have to worry about the plate sliding off very easily. If you’re the one who’s going to be printing this, I’d recommend that you print the puck upright and slice the arms so that they’re all separate. Once you download the stls, there’ll be a bit of post processing. for the arms especially, remember to use the split tool to separate them from each other. the .3mf file does this already, but you can also removed the grabber thing from three of the arms if you wish M3 screws and magnets recommended/needed Happy… Plating? Anyways, that’s it.", "repo": "https://github.com/vismaysjayaram/fieldPlate-CAD", "demo": ""}]...
|
[{"post_id": 50286, "ts": "202 [{"post_id": 50286, "ts": "2026-08-19T13:15:59Z", "hours": 2.1336, "text": "Devlog 2 Prototype 2 This is my second prototype. I looked at my feedback from the first one. The first prototype went well. There were definitely some slicing issues related to the Hilbert curve that I use on top. I don’t think my printer managed to fully capture how cool it looks, so I switched to a bit more of a simple pattern. I chose another, nicer geometric pattern. I added some screw holes, and I decided to design some arms. It’s like a four-piece arm with some screws, which allow it to stay in place. I added a little notch on one of the arms, which allows it to stay stable. I’ll be attaching some photos of P2. Testing I tested it, and in terms of stability, it’s much better. However, there were a couple of issues. Assembling it is still very annoying, simply because of the way I placed the screw holes. It was a little annoying to align the screws. I don’t think I made the chamfers to guide the screws in very nicely. I probably could have gotten away with larger holes on the arms, just so that they align a bit easier. I’ happy with how the arms came out. I probably could have gotten away with just a simple shell instead of all that fancy ribbing, but, you know, it’s okay. Slicing was definitely a pain, but I think I managed to get that done. It’s looking good, and I would make a P3 if I had the time, but unfortunately, I just don’t have the time. So this is the end of the project. Future Improvements In terms of the future, in terms of what I would love to do next, I got some much stronger 8 × 3 magnets, which I would like to use, and definitely figure out some better mounting system. Unfortunately, my summer vacation is over, so I don’t have the time. Also, I ran out of filament, so I couldn’t print the other three arms. Ciao!"}, {"post_id": 40647, "ts": "2026-08-04T10:19:53Z", "hours": 1.2833, "text": "Devlog 2 First Prototype This is my first prototype for fieldPlate.I’ll be attaching a picture of the prototype below, but I jumped into Fusion 360 and, based on my notes, came up with P1, or Prototype 1. There were a couple of things I did not foresee that I had to work around when prototyping. Surprisingly, the screws are rather big compared to the magnets that I’m using, so I had to come up with a weird cutout arrangement. When I started testing the prototype, I discovered that my magnets really aren’t all that strong. So that might be an issue that I have to fix by buying some new and stronger magnets for future prototypes. Testing In terms of stability, it’s okay. The small framework, especially the 42 × 42 frame, isn’t very stable. It’s just balancing at this point. However, it seems to work all right, and it’s honestly not too bad. I was able to hold a speaker directly offset from the frame so it was just hanging, and it managed to keep it in place. Next Steps Definitely, I’m going to have to work on the side mounts and also the puck which actually attaches to the screw holes.But that’s what I’m going to work on next.I’ll be attaching some pictures of what I built and what it’s looking like right now."}, {"post_id": 38806, "ts": "2026-08-01T11:25:23Z", "hours": 1.15, "text": "Devlog 1: Tricking a PEI Plate into holding Coffee The Idea All right, so this is the start of my new project, fieldPlate. fieldPlate is a funny little idea I had when sitting down to read a book. What if you attached a table onto a tripod?The premise is really very simple. I was thinking of ways that you could make a tripod more comfortable to rest your hand on for an armrest, as one does on a perfectly normal Saturday night. However, I realized that a PEI plate that you find on most common 3D printers is made out of metal, is magnetic, and is roughly the size of a table, which makes it quite portable. Thus I had the wonderful idea to try and create a project that could take a 3D printer printing plate, which is normally made out of metal, and convert that into a table which you can attach onto a tripod. And that’s basically what this project is going to be like. Prototyping In this session, all we did was jump into a notes app and spend about an hour prototyping by drawing a couple of ideas and designs.The final product will be about maybe 40 to 50 millimeters tall. It’ll be around 42 by 42 and 43 by 43 millimeters on both ends, respectively. The top end will be magnetic with a little bit of a dock or puck, which you can screw in. So if you have a different kind of magnet, you can remove the existing design and replace it with a remix of your own. (pls remix) There will also be four arms which connect to the corners of the field to prevent slipping or sheering. Finally, the arms can be screwed into place via three screws per side. Practicality Now, I will admit that this idea is not exactly the most practical, as most people don’t exactly go running around with extra PEI plates on their person. But I have a solution for that. I will also be releasing a plastic version of the plate so that you can simply screw it in and use it. Again, this idea is still not practical because there are more useful prototypes and products out there.But I think this is a funny idea, and it’s a good design challenge for me to try and apply some skills that I’ve been learning in Fusion 360 to this. I’ve never really worked with so many components before, and since this is a very hardware-focused product, and I already have a 3D printer, I think this will be a very interesting challenge for me. Comment any ideas pls. Maybe gridfinity?"}]...
|
Title: fieldPlate by @Maarxius | Stardance
Devlog Title: fieldPlate by @Maarxius | Stardance
Devlogs:
Devlog 1: Tricking a PEI Plate into holding Coffee The Idea All right, so this is the start of my new project, fieldPlate. fieldPlate is a funny little idea I had when sitting down to read a book. What if you attached a table onto a tripod?The premise is really very simple. I was thinking of ways that you could make a tripod more comfortable to rest your hand on for an armrest, as one does on a perfectly normal Saturday night. However, I realized that a PEI plate that you find on most common 3D printers is made out of metal, is magnetic, and is roughly the size of a table, which makes it quite portable. Thus I had the wonderful idea to try and create a project that could take a 3D printer printing plate, which is normally made out of metal, and convert that into a table which you can attach onto a tripod. And that’s basically what this project is going to be like. Prototyping In this session, all we did was jump into a notes app and spend about an hour prototyping by drawing a couple of ideas and designs.The final product will be about maybe 40 to 50 millimeters tall. It’ll be around 42 by 42 and 43 by 43 millimeters on both ends, respectively. The top end will be magnetic with a little bit of a dock or puck, which you can screw in. So if you have a different kind of magnet, you can remove the existing design and replace it with a remix of your own. (pls remix) There will also be four arms which connect to the corners of the field to prevent slipping or sheering. Finally, the arms can be screwed into place via three screws per side. Practicality Now, I will admit that this idea is not exactly the most practical, as most people don’t exactly go running around with extra PEI plates on their person. But I have a solution for that. I will also be releasing a plastic version of the plate so that you can simply screw it in and use it. Again, this idea is still not practical because there are more useful prototypes and products out there.But I think this is a funny idea, and it’s a good design challenge for me to try and apply some skills that I’ve been learning in Fusion 360 to this. I’ve never really worked with so many components before, and since this is a very hardware-focused product, and I already have a 3D printer, I think this will be a very interesting challenge for me. Comment any ideas pls. Maybe gridfinity?
Devlog 2 First Prototype This is my first prototype for fieldPlate.I’ll be attaching a picture of the prototype below, but I jumped into Fusion 360 and, based on my notes, came up with P1, or Prototype 1. There were a couple of things I did not foresee that I had to work around when prototyping. Surprisingly, the screws are rather big compared to the magnets that I’m using, so I had to come up with a weird cutout arrangement. When I started testing the prototype, I discovered that my magnets really aren’t all that strong. So that might be an issue that I have to fix by buying some new and stronger magnets for future prototypes. Testing In terms of stability, it’s okay. The small framework, especially the 42 × 42 frame, isn’t very stable. It’s just balancing at this point. However, it seems to work all right, and it’s honestly not too bad. I was able to hold a speaker directly offset from the frame so it was just hanging, and it managed to keep it in place. Next Steps Definitely, I’m going to have to work on the side mounts and also the puck which actually attaches to the screw holes.But that’s what I’m going to work on next.I’ll be attaching some pictures of what I built and what it’s looking like right now.
Devlog 2 Prototype 2 This is my second prototype. I looked at my feedback from the first one. The first prototype went well. There were definitely some slicing issues related to the Hilbert curve that I use on top. I don’t think my printer managed to fully capture how cool it looks, so I switched to a bit more of a simple pattern. I chose another, nicer geometric pattern. I added some screw holes, and I decided to design some arms. It’s like a four-piece arm with some screws, which allow it to stay in place. I added a little notch on one of the arms, which allows it to stay stable. I’ll be attaching some photos of P2. Testing I tested it, and in terms of stability, it’s much better. However, there were a couple of issues. Assembling it is still very annoying, simply because of the way I placed the screw holes. It was a little annoying to align the screws. I don’t think I made the chamfers to guide the screws in very nicely. I probably could have gotten away with larger holes on the arms, just so that they align a bit easier. I’ happy with how the arms came out. I probably could have gotten away with just a simple shell instead of all that fancy ribbing, but, you know, it’s okay. Slicing was definitely a pain, but I think I managed to get that done. It’s looking good, and I would make a P3 if I had the time, but unfortunately, I just don’t have the time. So this is the end of the project. Future Improvements In terms of the future, in terms of what I would love to do next, I got some much stronger 8 × 3 magnets, which I would like to use, and definitely figure out some better mounting system. Unfortunately, my summer vacation is over, so I don’t have the time. Also, I ran out of filament, so I couldn’t print the other three arms. Ciao!
Ships:
fieldPlate This project started out as a joke. I had a funny thought of converting a camera tripod into an armrest. Then I thought, I have a 3D printer plate that’s flat enough. Why not turn that into a table and stick it on top of a tripod? Maybe some godforsaken photographer could use it while they’re out in the field, and thus fieldPlate was born. Unfortunately, I do not have the time to do this project the full justice it deserves. That’s definitely a full P3 thing. However, I’m pretty proud of what I’ve done. In essence, it’s a very simple project. It’s just a tripod attachment with four arms, which allows you to hold things, and it works. It’s pretty good. It also has magnetic attachments, so you don’t have to worry about the plate sliding off very easily. If you’re the one who’s going to be printing this, I’d recommend that you print the puck upright and slice the arms so that they’re all separate. Once you download the stls, there’ll be a bit of post processing. for the arms especially, remember to use the split tool to separate them from each other. the .3mf file does this already, but you can also removed the grabber thing from three of the arms if you wish M3 screws and magnets recommended/needed Happy… Plating? Anyways, that’s it....
|
1793461666.496644
|
1793548066.496644
|
|
50245
|
my personal site by @mabu | Stardance
|
mabu
|
2
|
6.0
|
6
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T19:12:22Z", "hours": 4.0, "multiplier": 1.6, "payout": 6, "blessing": "neutral", "text": "This is my personal website i tried to make it look better than the first version and i have planned to do more to this later hope u guys like it", "repo": "https://github.com/mabu100/site-1-", "demo": "https://mabu100.github.io/site-1-/"}, {"ship_num": 2, "ts": "2026-08-17T16:32:59Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "This is a simple and ez website i made abt me i hope u guys like it and i want to add more in the future", "repo": "https://github.com/mabu100/site-1-", "demo": "https://mabu100.github.io/site-1-/"}]...
|
[{"post_id": 50550, "ts": "202 [{"post_id": 50550, "ts": "2026-08-19T19:11:12Z", "hours": 3.6133, "text": "Tried to make my personal site look better i have also added glow effect and contact will try to make it better in a few days"}, {"post_id": 48866, "ts": "2026-08-17T12:56:34Z", "hours": 1.5406, "text": "I just made a new version of my web page"}]...
|
Title: my personal site by @mabu | Stardance
Devl Title: my personal site by @mabu | Stardance
Devlogs:
I just made a new version of my web page
Tried to make my personal site look better i have also added glow effect and contact will try to make it better in a few days
Ships:
This is a simple and ez website i made abt me i hope u guys like it and i want to add more in the future
This is my personal website i tried to make it look better than the first version and i have planned to do more to this later hope u guys like it...
|
1792878529.371056
|
1793742529.371056
|
|
27619
|
skyOS by @skyFonix | Stardance
|
skyFonix
|
2
|
17.0
|
89
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T19:02:21Z", "hours": 9.0, "multiplier": 8.17, "payout": 89, "blessing": "neutral", "text": "Ship #2 Features added: Music player with metadata reading (cover art, title, artist, release year) Image viewer with zoom and grab to navigate, also supports setting images as wallpapers simpleShare integration, uploads, deletion, quota management and file opening with the appropriate media app, anything that is currently unsupported gets downloaded. simpleShare demo login info: username: demo password: demopass", "repo": "https://github.com/FonixPython/skyOS", "demo": "https://webos.cigoria.eu/"}, {"ship_num": 2, "ts": "2026-08-06T21:56:00Z", "hours": 8.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I made my very first webos! Features: About App, Calculator App, Notes App, Customizable wallpaper. The whole thing is written in React and hosted on my own hardware.", "repo": "https://github.com/FonixPython/skyOS", "demo": "https://webos.cigoria.eu/"}]...
|
[{"post_id": 50543, "ts": "202 [{"post_id": 50543, "ts": "2026-08-19T18:51:29Z", "hours": 1.4989, "text": "Added finishing touches to simpleShare integration, with a quota display, uploads, deletion and support for setting an uploaded image as the wallpaper!"}, {"post_id": 50317, "ts": "2026-08-19T14:12:00Z", "hours": 3.0203, "text": "Basics of simpleShare integration are complete, files can be opened inside skyOs directly from the files stored on simpleShare. I need to add uploads, deletion and wallpaper support to it."}, {"post_id": 49873, "ts": "2026-08-18T20:28:47Z", "hours": 2.2914, "text": "Added music player too, tried customizing the slider, but I couldn’t get it to react to css so I gave up ¯ ( ツ)/¯"}, {"post_id": 49433, "ts": "2026-08-18T08:34:30Z", "hours": 2.1956, "text": "Added Image viewer"}, {"post_id": 42208, "ts": "2026-08-06T20:24:47Z", "hours": 2.2136, "text": "Made a calculator too!"}, {"post_id": 41909, "ts": "2026-08-06T12:51:01Z", "hours": 1.7642, "text": "I made a notes app that saves to localStorage! I also added a wallpaper url to localStorage, but haven’t yet made the settings app"}, {"post_id": 41863, "ts": "2026-08-06T10:41:32Z", "hours": 1.1986, "text": "Added about app, and function to render different kinds of apps and close them!"}, {"post_id": 41837, "ts": "2026-08-06T09:26:18Z", "hours": 0.6336, "text": "Added draggable windows, and a button to open them!"}, {"post_id": 41823, "ts": "2026-08-06T08:45:16Z", "hours": 1.7706, "text": "Added top bar"}]...
|
Title: skyOS by @skyFonix | Stardance
Devlogs:
Ad Title: skyOS by @skyFonix | Stardance
Devlogs:
Added top bar
Added draggable windows, and a button to open them!
Added about app, and function to render different kinds of apps and close them!
I made a notes app that saves to localStorage! I also added a wallpaper url to localStorage, but haven’t yet made the settings app
Made a calculator too!
Added Image viewer
Added music player too, tried customizing the slider, but I couldn’t get it to react to css so I gave up ¯ ( ツ)/¯
Basics of simpleShare integration are complete, files can be opened inside skyOs directly from the files stored on simpleShare. I need to add uploads, deletion and wallpaper support to it.
Added finishing touches to simpleShare integration, with a quota display, uploads, deletion and support for setting an uploaded image as the wallpaper!
Ships:
I made my very first webos! Features: About App, Calculator App, Notes App, Customizable wallpaper. The whole thing is written in React and hosted on my own hardware.
Ship #2 Features added: Music player with metadata reading (cover art, title, artist, release year) Image viewer with zoom and grab to navigate, also supports setting images as wallpapers simpleShare integration, uploads, deletion, quota management and file opening with the appropriate media app, anything that is currently unsupported gets downloaded. simpleShare demo login info: username: demo password: demopass...
|
1793505225.9252012
|
1793591625.9252012
|
|
51307
|
My own site! by @Creatorr | Stardance
|
Creatorr
|
1
|
1.99
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T18:03:48Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "So I have build a personal website showcasing few of my projects which I am making during my learning journey. Website also contains things about me basically its a resume but in modern style. For me the minimalistic look was a challenge because I love minimal UI so I made sure to use it in this website and yeah it was a bit challenging like testing everything and the resolving the issue selecting the colour which aligns with the text and environment. There are lost of things which I have to add I will add those things when I will learn more and get to know about more things I will surely try to make it more and more amazing and cool. And I am proud of myself that instead of saying that ohh I don’t know this or this how will I build a website. I just dived into the coding process and here my website is live so yeah its quite exiting for me!", "repo": "https://github.com/Aryan-prog18/Personal-Website", "demo": "https://personal-website-wheat-nu.vercel.app"}]...
|
[{"post_id": 51200, "ts": "202 [{"post_id": 51200, "ts": "2026-08-20T18:31:57Z", "hours": 0.9903, "text": "So there was an error in my code and website: – The for the fade-in animation got removed while I was testing something and somehow I didn’t noticed that but now I fixed that so here are the things which I have changed in the code: 1. Restored the missing block before the - This is what reveals the .revel section. 2. Removed the extra tag in the about section. 3. Also I added a safety net a fallback plus a try/catch in the script, so even if somehow the JS ever fails again(blocked or any error), content shows by default instead of just disappearing. So yeah this was pretty much and I learned that sometimes a small bug can cause a big problem so we should always focus on everything and always double check the code before pushing it online or anywhere ."}, {"post_id": 50393, "ts": "2026-08-19T15:39:49Z", "hours": 0.8778, "text": "Well I am working on a personal projects and basically its a resume where I am showcasing projects which I have build so far"}]...
|
Title: My own site! by @Creatorr | Stardance
Devl Title: My own site! by @Creatorr | Stardance
Devlogs:
Well I am working on a personal projects and basically its a resume where I am showcasing projects which I have build so far
So there was an error in my code and website: – The for the fade-in animation got removed while I was testing something and somehow I didn’t noticed that but now I fixed that so here are the things which I have changed in the code: 1. Restored the missing block before the - This is what reveals the .revel section. 2. Removed the extra tag in the about section. 3. Also I added a safety net a fallback plus a try/catch in the script, so even if somehow the JS ever fails again(blocked or any error), content shows by default instead of just disappearing. So yeah this was pretty much and I learned that sometimes a small bug can cause a big problem so we should always focus on everything and always double check the code before pushing it online or anywhere .
Ships:
So I have build a personal website showcasing few of my projects which I am making during my learning journey. Website also contains things about me basically its a resume but in modern style. For me the minimalistic look was a challenge because I love minimal UI so I made sure to use it in this website and yeah it was a bit challenging like testing everything and the resolving the issue selecting the colour which aligns with the text and environment. There are lost of things which I have to add I will add those things when I will learn more and get to know about more things I will surely try to make it more and more amazing and cool. And I am proud of myself that instead of saying that ohh I don’t know this or this how will I build a website. I just dived into the coding process and here my website is live so yeah its quite exiting for me!...
|
1793486890.3328068
|
1793573290.3328068
|
|
30522
|
Lauren's Project Portfolio by @Lauren_hackalic Lauren's Project Portfolio by @Lauren_hackalicious | Stardance...
|
Lauren_hackalicious
|
1
|
23.0
|
297
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T17:41:42Z", "hours": 23.0, "multiplier": 13.05, "payout": 297, "blessing": "neutral", "text": "I created a project portfolio and personal website. You can learn a little about me as it pertains to my coding journey and involvement at school. I learned how to animate objects on screen and create variations of the same thing, for example, the banners. They all carry the same essence. And I learned how to create a slideshow effect. Users can swipe on the About Me page to view three different pictures relating to my after-school activities. I would like readers to know that they, depending on the size of their screen, zoom out by 25% for a proper viewing experience. This project was created on a MacBook Neo, so I had to zoom my screen out to program this website. I recommend users enable autoplay because this website automatically plays music on the landing and project pages. The last thing I would like to mention is to interact with my website! A lot of things have “overflow: scroll” enabled, so try to find everything that would require you to scroll. P.S. If you’re having a hard time navigating the website, click the Word , as it contains the href attributes.", "repo": "https://github.com/example0d1n/Portfolio", "demo": "https://example0d1n.github.io/Portfolio/"}]...
|
[{"post_id": 50485, "ts": "202 [{"post_id": 50485, "ts": "2026-08-19T17:27:32Z", "hours": 0.7914, "text": "Read.Me discretion. Zoom out your screen by 25% for a proper website viewing experience. If you don’t, unfortunately, elements will appear squished together. If I had more time this summer, I would create a more responsive layout. The picture linked to this devlog is the result of the website if it isn’t zoomed out enough for the proper viewing experience. I also had to replace all the videos on the project page with screenshots because the files were too large, over 100 MB, to push into a GitHub repository. This saddens me because it did make that page less static. PLEASE take the time to read the README."}, {"post_id": 50041, "ts": "2026-08-19T02:14:49Z", "hours": 3.9425, "text": "Website complete…for now I added music that plays automatically on the landing and projects page | I started and finished the About Me tab. It features a Linktree to all my socials. Beneath is a music player featuring my favorite songs from Olivia Rodrigo, Beabadoobee, KATSEYE, and Laufey. On the right, I briefly discuss my involvement at my school. I finished adding the captions to all my projects on the projects page. I decided to pause my progress on this website as school starts up for me, and I need to begin reviewing core subjects for my classes. I had fun creating this website and hope to continue working on it outside of Stardance."}, {"post_id": 48657, "ts": "2026-08-17T04:24:56Z", "hours": 1.6494, "text": "About Me update: I only began structuring the layout of how I want my Linktree to appear in this section of the website. The styling is NOWHERE near complete, especially since there will be a lot more I place under the Linktree to talk about what I do in school that relates to STEM/Computer science. The koi fish image is actually a gif"}, {"post_id": 47796, "ts": "2026-08-15T23:54:31Z", "hours": 3.0403, "text": "Project page complete for now!! I worked tirelessly to complete this page today because school starts soon, and I won’t have much time to work on any more huge web projects until my next seasonal break. I added 5 of my completed summer projects and included the portfolio website because it’ll be my biggest project yet! I recorded a live demo of each project that automatically plays upon opening the page, so users can see what each website is about without reading the bio. I also attached the live link just in case users want to try out the website for themselves. (I really wish I could attach a video to this devlog, but Stardance isn’t working with me when it comes to the videos"}, {"post_id": 47282, "ts": "2026-08-15T04:03:28Z", "hours": 1.4347, "text": "Landing page complete!: | aligned an ocean gif with the update logs to keep the layout consistent (balanced). I added a spinning dolphin because it was so funny to me and fit the Frutiger Aero theme anyway, so I was like… why not? I FINALLY updated the navigation buttons on the side because they were looking pretty plain compared to the rest of the page. In the next devlog, I will have worked on the project. The dolphin IS spinning"}, {"post_id": 46640, "ts": "2026-08-14T01:18:45Z", "hours": 0.6122, "text": "I’ve been having issues posting a devlog today due to this error. I’m going to edit this devlog once this stops appearing on my screen. I can’t find the screenshot of the original update to my website, but I did fix the unevenness of the pixel background widget and updated the weather widget! Check out my most recent devlog to see the changes; I’m having issues uploading videos to Stardance devlogs."}, {"post_id": 46115, "ts": "2026-08-13T03:36:37Z", "hours": 2.395, "text": "Center section complete…for now 1. I added a GIF with a scenic pixel background to lessen the static feel of the area! I matched the banner with the update log for consistency. 2. I moved on to the right side of the container. It took about 30 minutes to figure out how to place items on the right side because the way I programmed the container didn’t account for the layout. I went back and forth between moving divs around and asking GitHub Copilot how to fix the issue. I concluded that creating a separate div within the container div, outside the “main” div, resolved the issue. 3. After debugging, I added a live clock (in military time) and inserted a weather widget."}, {"post_id": 45506, "ts": "2026-08-12T01:27:56Z", "hours": 2.0222, "text": "Big big things todayyyy I finished customizing the welcome textbox by aligning the h1 tag to the left (horayy). I added an infinite carousel with things that I like/am interested in. The update log is primarily for future updates after I ship this portfolio and to keep up with the progress I make, even if it seems minimal. The Update log uses the CSS element “overflow: scroll”."}, {"post_id": 44992, "ts": "2026-08-11T03:55:44Z", "hours": 3.0992, "text": "Changed the theme again lol 😭 I made my portfolio website Frutiger Aero because it’s bright and easier for me to style!! I had to delete my original code since it made customizing the content inside the main container difficult! My fav thing about this version is the banner at the top of the main container."}, {"post_id": 44252, "ts": "2026-08-09T21:40:01Z", "hours": 2.3014, "text": "HUGE REVAMP I was inspired by early 2000s-themed websites and redesigned my portfolio to show more of my personality. The old version was too corporate for me. In the new design, I am incorporating themes of My Little Pony, one of my childhood shows! It is pretty bare-bones right now. To get the layout, I used Figma! I recommend this website as it really helps get the ball rolling. The colors are all over the place because I’m still trying to figure out how exactly I want everything placed. AKA, these are not the final colors/fonts/style."}, {"post_id": 21889, "ts": "2026-07-04T02:00:02Z", "hours": 1.44, "text": "1st Devlog I decided it was finally time to create a webpage dedicated to storing all my projects! As an aspiring computer science major, I feel this is crucial to have on deck. I played around with how I wanted things to be laid out and ended up with this design. Is it plain? Yes. Is it also straight to the point? Also yes. This will be amongst my shorter projects since all I need to do is fill out the template I created in HTML."}]...
|
Title: Lauren's Project Portfolio by @Lauren_h Title: Lauren's Project Portfolio by @Lauren_hackalicious | Stardance
Devlogs:
1st Devlog I decided it was finally time to create a webpage dedicated to storing all my projects! As an aspiring computer science major, I feel this is crucial to have on deck. I played around with how I wanted things to be laid out and ended up with this design. Is it plain? Yes. Is it also straight to the point? Also yes. This will be amongst my shorter projects since all I need to do is fill out the template I created in HTML.
HUGE REVAMP I was inspired by early 2000s-themed websites and redesigned my portfolio to show more of my personality. The old version was too corporate for me. In the new design, I am incorporating themes of My Little Pony, one of my childhood shows! It is pretty bare-bones right now. To get the layout, I used Figma! I recommend this website as it really helps get the ball rolling. The colors are all over the place because I’m still trying to figure out how exactly I want everything placed. AKA, these are not the final colors/fonts/style.
Changed the theme again lol 😭 I made my portfolio website Frutiger Aero because it’s bright and easier for me to style!! I had to delete my original code since it made customizing the content inside the main container difficult! My fav thing about this version is the banner at the top of the main container.
Big big things todayyyy I finished customizing the welcome textbox by aligning the h1 tag to the left (horayy). I added an infinite carousel with things that I like/am interested in. The update log is primarily for future updates after I ship this portfolio and to keep up with the progress I make, even if it seems minimal. The Update log uses the CSS element “overflow: scroll”.
Center section complete…for now 1. I added a GIF with a scenic pixel background to lessen the static feel of the area! I matched the banner with the update log for consistency. 2. I moved on to the right side of the container. It took about 30 minutes to figure out how to place items on the right side because the way I programmed the container didn’t account for the layout. I went back and forth between moving divs around and asking GitHub Copilot how to fix the issue. I concluded that creating a separate div within the container div, outside the “main” div, resolved the issue. 3. After debugging, I added a live clock (in military time) and inserted a weather widget.
I’ve been having issues posting a devlog today due to this error. I’m going to edit this devlog once this stops appearing on my screen. I can’t find the screenshot of the original update to my website, but I did fix the unevenness of the pixel background widget and updated the weather widget! Check out my most recent devlog to see the changes; I’m having issues uploading videos to Stardance devlogs.
Landing page complete!: | aligned an ocean gif with the update logs to keep the layout consistent (balanced). I added a spinning dolphin because it was so funny to me and fit the Frutiger Aero theme anyway, so I was like… why not? I FINALLY updated the navigation buttons on the side because they were looking pretty plain compared to the rest of the page. In the next devlog, I will have worked on the project. The dolphin IS spinning
Project page complete for now!! I worked tirelessly to complete this page today because school starts soon, and I won’t have much time to work on any more huge web projects until my next seasonal break. I added 5 of my completed summer projects and included the portfolio website because it’ll be my biggest project yet! I recorded a live demo of each project that automatically plays upon opening the page, so users can see what each website is about without reading the bio. I also attached the live link just in case users want to try out the website for themselves. (I really wish I could attach a video to this devlog, but Stardance isn’t working with me when it comes to the videos
About Me update: I only began structuring the layout of how I want my Linktree to appear in this section of the website. The styling is NOWHERE near complete, especially since there will be a lot more I place under the Linktree to talk about what I do in school that relates to STEM/Computer science. The koi fish image is actually a gif
Website complete…for now I added music that plays automatically on the landing and projects page | I started and finished the About Me tab. It features a Linktree to all my socials. Beneath is a music player featuring my favorite songs from Olivia Rodrigo, Beabadoobee, KATSEYE, and Laufey. On the right, I briefly discuss my involvement at my school. I finished adding the captions to all my projects on the projects page. I decided to pause my progress on this website as school starts up for me, and I need to begin reviewing core subjects for my classes. I had fun creating this website and hope to continue working on it outside of Stardance.
Read.Me discretion. Zoom out your screen by 25% for a proper website viewing experience. If you don’t, unfortunately, elements will appear squished together. If I had more time this summer, I would create a more responsive layout. The picture linked to this devlog is the result of the website if it isn’t zoomed out enough for the proper viewing experience. I also had to replace all the videos on the project page with screenshots because the files were too large, over 100 MB, to push into a GitHub repository. This saddens me because it did make that page less static. PLEASE take the time to read the README.
Ships:
I created a project portfolio and personal website. You can learn a little about me as it pertains to my coding journey and involvement at school. I learned how to animate objects on screen and create variations of the same thing, for example, the banners. They all carry the same essence. And I learned how to create a slideshow effect. Users can swipe on the About Me page to view three different pictures relating to my after-school activities. I would like readers to know that they, depending on the size of their screen, zoom out by 25% for a proper viewing experience. This project was created on a MacBook Neo, so I had to zoom my screen out to program this website. I recommend users enable autoplay because this website automatically plays music on the landing and project pages. The last thing I would like to mention is to interact with my website! A lot of things have “overflow: scroll” enabled, so try to find everything that would require you to scroll. P.S. If you’re having a hard time navigating the website, click the Word , as it contains the href attributes....
|
1793325343.3518615
|
1793411743.3518615
|
|
50859
|
Escape your Fate by @Zincx_high | Stardance
|
Zincx_high
|
1
|
5.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T17:16:44Z", "hours": 5.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I made a horror datapack inspired by dead by daylight,so one day I was plating minecarft and I was on a pretty big winstreak nobody was able to win from me but after dominating soo much I got board and went to youtube I saw ppl playing horror games with friends and I thougth “wouldn’t it be cool to play a horror game on minecraft” my brain started to think and said “ Man it would be so cool if I made it myself, how hard could it be” and then I met the horror of proper directory managment for code to work took me a day to figure where each file goes and then I started lerning mcfuncation(yes its programming code I checked it) and I tougth shit its just programming and then I started cooking to the point I burned it and the game started having Immence lag I didn’t know how to fix it and I turned towards ai humanities enemy and I remmebered that humility is needed in order to suck seed so I sowlled my pride and gave in to fix the big and made this master piece come to life and yeah thats it.", "repo": "", "demo": ""}]...
|
[{"post_id": 50451, "ts": "202 [{"post_id": 50451, "ts": "2026-08-19T16:46:42Z", "hours": 5.0722, "text": "Devlog: I already created everything in the datapack will it might not be the final result but the gameplay loop is established and the story is done you can hope on and play but I have many more tasks and things like camera direction to put in this pack in the future rn enjoy"}]...
|
Title: Escape your Fate by @Zincx_high | Stardance Title: Escape your Fate by @Zincx_high | Stardance
Devlogs:
Devlog: I already created everything in the datapack will it might not be the final result but the gameplay loop is established and the story is done you can hope on and play but I have many more tasks and things like camera direction to put in this pack in the future rn enjoy
Ships:
I made a horror datapack inspired by dead by daylight,so one day I was plating minecarft and I was on a pretty big winstreak nobody was able to win from me but after dominating soo much I got board and went to youtube I saw ppl playing horror games with friends and I thougth “wouldn’t it be cool to play a horror game on minecraft” my brain started to think and said “ Man it would be so cool if I made it myself, how hard could it be” and then I met the horror of proper directory managment for code to work took me a day to figure where each file goes and then I started lerning mcfuncation(yes its programming code I checked it) and I tougth shit its just programming and then I started cooking to the point I burned it and the game started having Immence lag I didn’t know how to fix it and I turned towards ai humanities enemy and I remmebered that humility is needed in order to suck seed so I sowlled my pride and gave in to fix the big and made this master piece come to life and yeah thats it....
|
1793407723.9770243
|
1793494123.9770243
|
|
51466
|
BioStack by @onlybugs05 | Stardance
|
onlybugs05
|
1
|
1.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T16:45:50Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I built BioStack Under 1 Hour", "repo": "https://github.com/onlybugs05/BioStack/blob/main", "demo": "https://onlybugs05.github.io/BioStack/"}]...
|
[{"post_id": 50447, "ts": "202 [{"post_id": 50447, "ts": "2026-08-19T16:42:36Z", "hours": 0.6686, "text": "Done Building It"}]...
|
Title: BioStack by @onlybugs05 | Stardance
Devlog Title: BioStack by @onlybugs05 | Stardance
Devlogs:
Done Building It
Ships:
I built BioStack Under 1 Hour...
|
1793501480.2901852
|
1793587880.2901852
|
|
51386
|
jsnap.xyz redesign 4 by @JSnap | Stardance
|
JSnap
|
1
|
2.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T16:32:23Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I finally got around to re-designing my website. The old one left a lot to be desired. The first page is a simple splash screen, with my name- and my main social media buttons. However the “see more” or the second page is where it gets interesting. This page has a sticky note-style mosaic, each filled with individual little notes. The first section explores my setup, whereas the second section is going to be used as a mini-blog. AI was used within the CSS to allow me to optimise for smaller screens. I wrote the original code, and then the AI did the optimisations", "repo": "https://github.com/JSnapXYZ/website", "demo": "https://jsnap.xyz"}]...
|
[{"post_id": 50414, "ts": "202 [{"post_id": 50414, "ts": "2026-08-19T16:08:11Z", "hours": 2.4294, "text": "i did everything. hello all! I completely re-designed my website from the ground up."}]...
|
Title: jsnap.xyz redesign 4 by @JSnap | Stardance
Title: jsnap.xyz redesign 4 by @JSnap | Stardance
Devlogs:
i did everything. hello all! I completely re-designed my website from the ground up.
Ships:
I finally got around to re-designing my website. The old one left a lot to be desired. The first page is a simple splash screen, with my name- and my main social media buttons. However the “see more” or the second page is where it gets interesting. This page has a sticky note-style mosaic, each filled with individual little notes. The first section explores my setup, whereas the second section is going to be used as a mini-blog. AI was used within the CSS to allow me to optimise for smaller screens. I wrote the original code, and then the AI did the optimisations...
|
1793490630.6725225
|
1793577030.6725225
|
|
50113
|
Automated Student Performance Analyzer (Python + P Automated Student Performance Analyzer (Python + Pandas) by @govindrkumar | Stardance...
|
govindrkumar
|
1
|
10.0
|
48
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T16:18:13Z", "hours": 10.0, "multiplier": 4.62, "payout": 48, "blessing": "neutral", "text": "I made a web app that helps automatically analyze student marks. You can either enter the marks manually or upload a CSV/XLSX file. The app then calculates total marks, finds subject-wise toppers, and creates graphs to show performance. The challenging part was making both input methods work with the same analysis system and fixing the different data formats. I’m proud that I was able to build and deploy the whole project myself. To test it, you can either enter some student marks manually or upload a CSV/XLSX file and check the generated results and graphs. Try it : https://automated-student-performance-analyzer.onrender.com", "repo": "https://github.com/govindrkumar/Automated-Student-Performance-Analyzer", "demo": "https://automated-student-performance-analy.vercel.app/"}]...
|
[{"post_id": 50419, "ts": "202 [{"post_id": 50419, "ts": "2026-08-19T16:13:50Z", "hours": 0.3108, "text": "Manual Working done! Home page redesigned too (Added a title in form and colored it brown. Nothing much!). I am not very good in css. So, sorry for leaving it with weak design. My all 4 QoLs are finished with this project. I didn’t see, use of MySQL so I had to drop it. For anyone trying to contribute is welcome for PR. Project try link : https://automated-student-performance-analyzer.onrender.com"}, {"post_id": 50401, "ts": "2026-08-19T15:46:52Z", "hours": 4.6431, "text": "So, In last two days : I successfully implemented my intended architecture. Firstly, I successfully implemented automatic upload of csv or xls file and analysis of dataframe. It provides total marks of all students, each subject toppers and graphical comparison of all students. Today, I woke up and started working on Manual upload. So, when user manually uploads data, they get a language selection panel. Afterward, according to their selection a table is generated. From where, they can enter the data and get same functionality which they were getting while automatic upload. So, now I have to correct some style.css and it will be ready for final shipment tomorrow."}, {"post_id": 49406, "ts": "2026-08-18T06:55:30Z", "hours": 5.3747, "text": "I just made a student result analyzer. Here, you can upload your .csv or .xlsx file and analyze your student marks, get the marks statement of everyone, topper list, subject-wise performance of each student using matplotlib graph. It uses flask in backend."}]...
|
Title: Automated Student Performance Analyzer (Pyt Title: Automated Student Performance Analyzer (Python + Pandas) by @govindrkumar | Stardance
Devlogs:
I just made a student result analyzer. Here, you can upload your .csv or .xlsx file and analyze your student marks, get the marks statement of everyone, topper list, subject-wise performance of each student using matplotlib graph. It uses flask in backend.
So, In last two days : I successfully implemented my intended architecture. Firstly, I successfully implemented automatic upload of csv or xls file and analysis of dataframe. It provides total marks of all students, each subject toppers and graphical comparison of all students. Today, I woke up and started working on Manual upload. So, when user manually uploads data, they get a language selection panel. Afterward, according to their selection a table is generated. From where, they can enter the data and get same functionality which they were getting while automatic upload. So, now I have to correct some style.css and it will be ready for final shipment tomorrow.
Manual Working done! Home page redesigned too (Added a title in form and colored it brown. Nothing much!). I am not very good in css. So, sorry for leaving it with weak design. My all 4 QoLs are finished with this project. I didn’t see, use of MySQL so I had to drop it. For anyone trying to contribute is welcome for PR. Project try link : https://automated-student-performance-analyzer.onrender.com
Ships:
I made a web app that helps automatically analyze student marks. You can either enter the marks manually or upload a CSV/XLSX file. The app then calculates total marks, finds subject-wise toppers, and creates graphs to show performance. The challenging part was making both input methods work with the same analysis system and fixing the different data formats. I’m proud that I was able to build and deploy the whole project myself. To test it, you can either enter some student marks manually or upload a CSV/XLSX file and check the generated results and graphs. Try it : https://automated-student-performance-analyzer.onrender.com...
|
1793497694.8020055
|
1793584094.8020055
|
|
48897
|
mubeeeel by @mohamed_mubeel_s | Stardance
|
mohamed_mubeel_s
|
1
|
11.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T16:12:30Z", "hours": 11.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Building NeuroSift: Edge-First AST Engine, Hybrid Code Intelligence & 3D Knowledge Graph Just wrapped up building and open-sourcing NeuroSift — a local, privacy-first semantic code search daemon and interactive 3D dependency visualizer. When navigating large codebases, traditional string search is too noisy, and sending proprietary code to third-party cloud LLMs creates latency and privacy risks. NeuroSift solves this completely on-device. What NeuroSift does under the hood: Static AST & Complexity Auditing: Traverses Python abstract syntax trees to compute Cyclomatic Complexity and Halstead metrics while detecting code smells in real time. Knowledge Graph & Blast Radius: Builds directed dependency graphs mapping function calls, module imports, and inheritance trees to instantly calculate downstream impact across the codebase. Hybrid Search (Dense + BM25): Combines 384-dimensional vector embeddings with an exact-token BM25 lexical engine using Reciprocal Rank Fusion (RRF) for sub-15ms code retrieval. Dual Interface (TUI & WebGL 3D Visualizer): Features a terminal dashboard and a browser-based D3/WebGL force-directed visualizer with code previews and live telemetry. Zero Cloud Dependency: Runs 100% locally with 286+ unit tests passing cleanly.", "repo": "https://github.com/mohamedmubeels/Neurosift", "demo": "https://drive.google.com/file/d/1OlAwdFJmfrGDkMcbTguypytR31jB2sqj/view?usp=sharing"}]...
|
[{"post_id": 50356, "ts": "202 [{"post_id": 50356, "ts": "2026-08-19T15:09:20Z", "hours": 0.8806, "text": "Building NeuroSift: Edge-First AST Engine, Hybrid Code Intelligence & 3D Knowledge GraphJust wrapped up building and open-sourcing NeuroSift — a local, privacy-first semantic code search daemon and interactive 3D dependency visualizer.When navigating large codebases, traditional string search is too noisy, and sending proprietary code to third-party cloud LLMs creates latency and privacy risks. NeuroSift solves this completely on-device.What NeuroSift does under the hood:Static AST & Complexity Auditing: Traverses Python abstract syntax trees to compute Cyclomatic Complexity and Halstead metrics while detecting code smells in real time.Knowledge Graph & Blast Radius: Builds directed dependency graphs mapping function calls, module imports, and inheritance trees to instantly calculate downstream impact across the codebase.Hybrid Search (Dense + BM25): Combines 384-dimensional vector embeddings with an exact-token BM25 lexical engine using Reciprocal Rank Fusion (RRF) for sub-15ms code retrieval.Dual Interface (TUI & WebGL 3D Visualizer): Features a terminal dashboard and a browser-based D3/WebGL force-directed visualizer with code previews and live telemetry.Zero Cloud Dependency: Runs 100% locally with 286+ unit tests passing cleanly."}, {"post_id": 49703, "ts": "2026-08-18T16:34:10Z", "hours": 10.1747, "text": "Just wrapped up an intensive 8-hour engineering block building NeuroSift — a local, privacy-first semantic code search engine and dependency graph visualizer.Here is the breakdown of what got built and validated across Hours 1 through 8:Hours 1–2 (AST Parsing & Knowledge Graph): Built a static analysis engine with Python’s ast module (ast_extractor.py) to extract function/class definitions, calculate Cyclomatic Complexity & Halstead metrics, and flag code smells. Implemented builder.py to construct directed dependency graphs, identify caller-callee hierarchies, detect circular cycles, and calculate downstream impact “blast radius”.Hours 3–4 (Semantic Chunking & Local Vector Store): Created chunker.py to slice code by semantic function/class boundaries while injecting parameter signatures and docstring preambles. Developed embedder.py with SHA-256 disk caching and built an in-memory NumPy vector database with Cosine Similarity ranking.Hours 5–6 (BM25 Lexical & Terminal TUI): Built an exact-token BM25 sparse index and unified it with dense vectors via Reciprocal Rank Fusion (RRF). Developed a styled Rich terminal TUI with real-time complexity health badges, interactive search loops, and ASCII blast-radius dependency trees.Hours 7–8 (Daemon & Interactive Web Visualizer): Deployed a local REST/WebSocket daemon with FastAPI (and fallback standard library server) and connected it to a full-viewport D3.js force-directed visualizer, code preview drawer, and live telemetry stream.Stats: 286/286 unit tests passing in ~4.2s. 133 chunks indexed across the engine."}]...
|
Title: mubeeeel by @mohamed_mubeel_s | Stardance
Title: mubeeeel by @mohamed_mubeel_s | Stardance
Devlogs:
Just wrapped up an intensive 8-hour engineering block building NeuroSift — a local, privacy-first semantic code search engine and dependency graph visualizer.Here is the breakdown of what got built and validated across Hours 1 through 8:Hours 1–2 (AST Parsing & Knowledge Graph): Built a static analysis engine with Python’s ast module (ast_extractor.py) to extract function/class definitions, calculate Cyclomatic Complexity & Halstead metrics, and flag code smells. Implemented builder.py to construct directed dependency graphs, identify caller-callee hierarchies, detect circular cycles, and calculate downstream impact “blast radius”.Hours 3–4 (Semantic Chunking & Local Vector Store): Created chunker.py to slice code by semantic function/class boundaries while injecting parameter signatures and docstring preambles. Developed embedder.py with SHA-256 disk caching and built an in-memory NumPy vector database with Cosine Similarity ranking.Hours 5–6 (BM25 Lexical & Terminal TUI): Built an exact-token BM25 sparse index and unified it with dense vectors via Reciprocal Rank Fusion (RRF). Developed a styled Rich terminal TUI with real-time complexity health badges, interactive search loops, and ASCII blast-radius dependency trees.Hours 7–8 (Daemon & Interactive Web Visualizer): Deployed a local REST/WebSocket daemon with FastAPI (and fallback standard library server) and connected it to a full-viewport D3.js force-directed visualizer, code preview drawer, and live telemetry stream.Stats: 286/286 unit tests passing in ~4.2s. 133 chunks indexed across the engine.
Building NeuroSift: Edge-First AST Engine, Hybrid Code Intelligence & 3D Knowledge GraphJust wrapped up building and open-sourcing NeuroSift — a local, privacy-first semantic code search daemon and interactive 3D dependency visualizer.When navigating large codebases, traditional string search is too noisy, and sending proprietary code to third-party cloud LLMs creates latency and privacy risks. NeuroSift solves this completely on-device.What NeuroSift does under the hood:Static AST & Complexity Auditing: Traverses Python abstract syntax trees to compute Cyclomatic Complexity and Halstead metrics while detecting code smells in real time.Knowledge Graph & Blast Radius: Builds directed dependency graphs mapping function calls, module imports, and inheritance trees to instantly calculate downstream impact across the codebase.Hybrid Search (Dense + BM25): Combines 384-dimensional vector embeddings with an exact-token BM25 lexical engine using Reciprocal Rank Fusion (RRF) for sub-15ms code retrieval.Dual Interface (TUI & WebGL 3D Visualizer): Features a terminal dashboard and a browser-based D3/WebGL force-directed visualizer with code previews and live telemetry.Zero Cloud Dependency: Runs 100% locally with 286+ unit tests passing cleanly.
Ships:
Building NeuroSift: Edge-First AST Engine, Hybrid Code Intelligence & 3D Knowledge Graph Just wrapped up building and open-sourcing NeuroSift — a local, privacy-first semantic code search daemon and interactive 3D dependency visualizer. When navigating large codebases, traditional string search is too noisy, and sending proprietary code to third-party cloud LLMs creates latency and privacy risks. NeuroSift solves this completely on-device. What NeuroSift does under the hood: Static AST & Complexity Auditing: Traverses Python abstract syntax trees to compute Cyclomatic Complexity and Halstead metrics while detecting code smells in real time. Knowledge Graph & Blast Radius: Builds directed dependency graphs mapping function calls, module imports, and inheritance trees to instantly calculate downstream impact across the codebase. Hybrid Search (Dense + BM25): Combines 384-dimensional vector embeddings with an exact-token BM25 lexical engine using Reciprocal Rank Fusion (RRF) for sub-15ms code retrieval. Dual Interface (TUI & WebGL 3D Visualizer): Features a terminal dashboard and a browser-based D3/WebGL force-directed visualizer with code previews and live telemetry. Zero Cloud Dependency: Runs 100% locally with 286+ unit tests passing cleanly....
|
1793498089.199492
|
1793584489.199492
|
|
45740
|
Statitron by @matest | Stardance
|
matest
|
1
|
18.53
|
197
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T15:15:20Z", "hours": 17.0, "multiplier": 11.58, "payout": 197, "blessing": "neutral", "text": "Statitron is a hardware monitor that tracks: GPU Usage in percent, GPU temp, GPU Vram usage in percent CPU Usage RAM Usage in both percent and GB Net Recv + Net Send All disk partitions usage in GB and percent + Disk write and disk read total from all drives It also has a cool feature which is event tracker. It tracks GPU, CPU and RAM usage and whenever one of them has a spike of 20%, it records it and displays it in the window. Every 5 seconds it will also display updates on how your hardware usage is at that moment. Through this project i learnt 5 new python libraries that i before only had minimal experience with! This is the first ever project i made that id say im proud of, because it taught me so much and i had a lot of fun making it. Also the first time my ui design looks at least semi-decent :)", "repo": "https://github.com/TheMathematico/Statitron1", "demo": "https://github.com/TheMathematico/Statitron1/releases/tag/v1.0.0"}]...
|
[{"post_id": 57442, "ts": "202 [{"post_id": 57442, "ts": "2026-08-29T16:23:25Z", "hours": 1.5261, "text": "accidently posted a devlog"}, {"post_id": 47190, "ts": "2026-08-14T22:28:55Z", "hours": 6.1078, "text": "Final devlog before shipping(#4) It has been 2 days since my last devlog, here’s what happened: I’ve finished the event tracker, it shows cpu, gpu and ram spikes + usage updates every few seconds. You can also snooze updates or snooze both the spikes and the updates(snooze=they dont appear in the event tracker). Before shipping(tommorow) i will probably organize the code a little bit and maybe add a bar that shows the health of all the disks you have on your pc. Im glad that im finally done and i hope it looks good! (in the vid the event tracker is mistakenly named gpu monitor, sry)"}, {"post_id": 45445, "ts": "2026-08-11T21:56:28Z", "hours": 4.2778, "text": "Devlog - 3 Today i’ve added a detailed monitor for disk, gpu and ram. The disk monitor shows write speed and read speed combined from all the drives you have on your pc, and it also shows used and total space from all partitions you have(Don’t ask why my partitions are so full lol). The GPU monitor shows temp, usage and vram utilization. The Ram monitor is pretty basic and just shows how much is being utilized in gb and percent. For the next devlog im planning on importing another library to track your disk health and finally start adding the event tracker(a tracker that shows what app caused what component spike eg. google chrome caused a cpu spike at some point here’s what you can do etc.) This program also made me realize that my drive fills up passively every hour even tho im not installing anything which is so cool :)"}, {"post_id": 44809, "ts": "2026-08-10T20:42:31Z", "hours": 4.6183, "text": "Devlog - 2 I completely reworked the code and added 6 cool looking meters that show cpu/gpu usage, disk space, ram used and net recv + net send(net download/upload). I haven’t added the event tracker yet but im hoping to tomorrow. Im not quite sure what to put in the empty space below the buttons so i’d love to hear some suggestions!"}, {"post_id": 44285, "ts": "2026-08-09T22:51:32Z", "hours": 2.0314, "text": "Devlog - 1: Coded the stats monitor which i spent like 30 mins actually coding and 1h 30 mins cleaning up my messy code. For now, it tracks cpu usage, ram and gpu. Next i’ll be polishing the UI which doesn’t look the best at the moment and adding an event tracker that tracks hardware usage spikes and what process/app caused it."}]...
|
Title: Statitron by @matest | Stardance
Devlogs:
Title: Statitron by @matest | Stardance
Devlogs:
Devlog - 1: Coded the stats monitor which i spent like 30 mins actually coding and 1h 30 mins cleaning up my messy code. For now, it tracks cpu usage, ram and gpu. Next i’ll be polishing the UI which doesn’t look the best at the moment and adding an event tracker that tracks hardware usage spikes and what process/app caused it.
Devlog - 2 I completely reworked the code and added 6 cool looking meters that show cpu/gpu usage, disk space, ram used and net recv + net send(net download/upload). I haven’t added the event tracker yet but im hoping to tomorrow. Im not quite sure what to put in the empty space below the buttons so i’d love to hear some suggestions!
Devlog - 3 Today i’ve added a detailed monitor for disk, gpu and ram. The disk monitor shows write speed and read speed combined from all the drives you have on your pc, and it also shows used and total space from all partitions you have(Don’t ask why my partitions are so full lol). The GPU monitor shows temp, usage and vram utilization. The Ram monitor is pretty basic and just shows how much is being utilized in gb and percent. For the next devlog im planning on importing another library to track your disk health and finally start adding the event tracker(a tracker that shows what app caused what component spike eg. google chrome caused a cpu spike at some point here’s what you can do etc.) This program also made me realize that my drive fills up passively every hour even tho im not installing anything which is so cool :)
Final devlog before shipping(#4) It has been 2 days since my last devlog, here’s what happened: I’ve finished the event tracker, it shows cpu, gpu and ram spikes + usage updates every few seconds. You can also snooze updates or snooze both the spikes and the updates(snooze=they dont appear in the event tracker). Before shipping(tommorow) i will probably organize the code a little bit and maybe add a bar that shows the health of all the disks you have on your pc. Im glad that im finally done and i hope it looks good! (in the vid the event tracker is mistakenly named gpu monitor, sry)
accidently posted a devlog
Ships:
Statitron is a hardware monitor that tracks: GPU Usage in percent, GPU temp, GPU Vram usage in percent CPU Usage RAM Usage in both percent and GB Net Recv + Net Send All disk partitions usage in GB and percent + Disk write and disk read total from all drives It also has a cool feature which is event tracker. It tracks GPU, CPU and RAM usage and whenever one of them has a spike of 20%, it records it and displays it in the window. Every 5 seconds it will also display updates on how your hardware usage is at that moment. Through this project i learnt 5 new python libraries that i before only had minimal experience with! This is the first ever project i made that id say im proud of, because it taught me so much and i had a lot of fun making it. Also the first time my ui design looks at least semi-decent :)...
|
1793541255.858976
|
1793627655.858976
|
|
45886
|
DeskPet by @Cherish9 | Stardance
|
Cherish9
|
1
|
7.08
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T14:27:53Z", "hours": 4.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Building a smart “DeskPet” companion with Arduino & Python! It tracks my daily water intake, monitors room temperature and humidity to tell me when to open a window, and keeps my desk healthy.", "repo": "https://github.com/Cherish-everything/DeskPet-A-well-being-assistant", "demo": ""}]...
|
[{"post_id": 76106, "ts": "202 [{"post_id": 76106, "ts": "2026-10-02T10:22:33Z", "hours": 1.9503, "text": "Build an enclosure on freecad! Honestly it was my first time building an enclosure with the pcb model, it wasn’t too bad actually. And I’m getting used to cad now. My next update would be probably cleaning up my code and organizing my repository for this"}, {"post_id": 75830, "ts": "2026-10-01T08:53:01Z", "hours": 1.1331, "text": "I’m working on a schematic for deskpet, this is like my second PCB design project, next session I will work on the pcb itself!"}, {"post_id": 50326, "ts": "2026-08-19T14:26:25Z", "hours": 0.31, "text": "Just logging the time I spent debugging a bit and documenting the project for the demo video"}, {"post_id": 47404, "ts": "2026-08-15T10:00:45Z", "hours": 1.3736, "text": "I’m finalizing my deskpet v1: just a simple working prototype since this project is intended as a learning project so I won’t expand on it too much. Today I got it to take water intake logs through a button you press and it reminds you to open a window or drink water while you’re working. Initially wanted to add a screen but I ran out of jumper wires:/ Flipping hell I missspelled deskpet on lapse so the hardware part isn’t recorded aka about 15 minutes of me debugging a button only to find out it was faulty"}, {"post_id": 44463, "ts": "2026-08-10T07:13:46Z", "hours": 1.0239, "text": "Tried adding music to the buzzer lol and added a water sensor to remind to drink"}, {"post_id": 43814, "ts": "2026-08-09T07:47:40Z", "hours": 0.9183, "text": "The code + prototype circuit. My first time using these sensors for an actual project!"}]...
|
Title: DeskPet by @Cherish9 | Stardance
Devlogs:
Title: DeskPet by @Cherish9 | Stardance
Devlogs:
The code + prototype circuit. My first time using these sensors for an actual project!
Tried adding music to the buzzer lol and added a water sensor to remind to drink
I’m finalizing my deskpet v1: just a simple working prototype since this project is intended as a learning project so I won’t expand on it too much. Today I got it to take water intake logs through a button you press and it reminds you to open a window or drink water while you’re working. Initially wanted to add a screen but I ran out of jumper wires:/ Flipping hell I missspelled deskpet on lapse so the hardware part isn’t recorded aka about 15 minutes of me debugging a button only to find out it was faulty
Just logging the time I spent debugging a bit and documenting the project for the demo video
I’m working on a schematic for deskpet, this is like my second PCB design project, next session I will work on the pcb itself!
Build an enclosure on freecad! Honestly it was my first time building an enclosure with the pcb model, it wasn’t too bad actually. And I’m getting used to cad now. My next update would be probably cleaning up my code and organizing my repository for this
Ships:
Building a smart “DeskPet” companion with Arduino & Python! It tracks my daily water intake, monitors room temperature and humidity to tell me when to open a window, and keeps my desk healthy....
|
1793566429.318828
|
1793652829.318828
|
|
49376
|
My own site! by @lmao_DK | Stardance
|
lmao_DK
|
2
|
3.0
|
1
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T13:41:07Z", "hours": 0.0, "multiplier": 2.13, "payout": 1, "blessing": "neutral", "text": "just finished with all the updates", "repo": "https://github.com/Code-Breaker16/personal-site", "demo": "https://personal-site-eta-ebon.vercel.app/"}, {"ship_num": 2, "ts": "2026-08-17T18:22:42Z", "hours": 3.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I just completed my first project here not my first html project but one of the complex yet easy ones ive worked on. The java script part a bit tricky but im glad i pushed myself to acheive it and finally i love how the whole site comes out many of u might thinks its ai generated but trust me it is not ive worked 3hrs on this basic project to make it look how it looks rn. Thank you and hope u guys will show some love to it .", "repo": "https://github.com/Code-Breaker16/personal-site", "demo": "https://personal-site-eta-ebon.vercel.app/"}]...
|
[{"post_id": 50278, "ts": "202 [{"post_id": 50278, "ts": "2026-08-19T13:04:28Z", "hours": 0.3797, "text": "just finished with all the updates"}, {"post_id": 49083, "ts": "2026-08-17T18:14:50Z", "hours": 1.7508, "text": "finally done with the project"}, {"post_id": 48365, "ts": "2026-08-16T17:57:15Z", "hours": 0.6386, "text": "Just done with the main html part starting off with css and jss"}, {"post_id": 48038, "ts": "2026-08-16T10:17:32Z", "hours": 0.4811, "text": "Just started working on my site not my first website, but the most special one. Done with the first section, the About Me part. Now I’ll be taking a break and studying for my exam tomorrow"}]...
|
Title: My own site! by @lmao_DK | Stardance
Devlo Title: My own site! by @lmao_DK | Stardance
Devlogs:
Just started working on my site not my first website, but the most special one. Done with the first section, the About Me part. Now I’ll be taking a break and studying for my exam tomorrow
Just done with the main html part starting off with css and jss
finally done with the project
just finished with all the updates
Ships:
I just completed my first project here not my first html project but one of the complex yet easy ones ive worked on. The java script part a bit tricky but im glad i pushed myself to acheive it and finally i love how the whole site comes out many of u might thinks its ai generated but trust me it is not ive worked 3hrs on this basic project to make it look how it looks rn. Thank you and hope u guys will show some love to it .
just finished with all the updates...
|
1793551838.4672477
|
1793638238.4672477
|
|
50777
|
rustymusty(rm) tab by @prakit | Stardance
|
|
1
|
3.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T13:14:32Z", "hours": 3.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "rusty musty is like a home page for a browser ( supports two search engines - google and duckduckgo [ i like duckduckgo ] ) . simple ui so that its easy to navigate , users can authenticate themeselves to the tab page but its completely optional", "repo": "https://github.com/zenformality/rustymusty", "demo": "https://rustymusty-vy1r.vercel.app/"}]...
|
[{"post_id": 50271, "ts": "202 [{"post_id": 50271, "ts": "2026-08-19T13:01:24Z", "hours": 0.565, "text": "Added discord authentication to rusty musty and now its complete!!!! YAYAYY"}, {"post_id": 49688, "ts": "2026-08-18T16:06:51Z", "hours": 0.4214, "text": "fixed alot of bugs , switched to a select element to make it easier to switch between google and duckduckgo"}, {"post_id": 49617, "ts": "2026-08-18T14:35:35Z", "hours": 1.5394, "text": "It works! real mess ngl"}]...
|
Title: rustymusty(rm) tab by @prakit | Stardance
Title: rustymusty(rm) tab by @prakit | Stardance
Devlogs:
It works! real mess ngl
fixed alot of bugs , switched to a select element to make it easier to switch between google and duckduckgo
Added discord authentication to rusty musty and now its complete!!!! YAYAYY
Ships:
rusty musty is like a home page for a browser ( supports two search engines - google and duckduckgo [ i like duckduckgo ] ) . simple ui so that its easy to navigate , users can authenticate themeselves to the tab page but its completely optional...
|
1792878836.366966
|
1793742836.366966
|
|
50976
|
Personal Site by @Acidicts | Stardance
|
Acidicts
|
1
|
4.39
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T12:56:53Z", "hours": 4.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Made the whole Site: in 3 sections seen as etc to 3 with footer and some api usage for live hackatime info", "repo": "https://github.com/Acidicts/personal-site", "demo": "https://acidicts.github.io/personal-site/"}]...
|
[{"post_id": 51018, "ts": "202 [{"post_id": 51018, "ts": "2026-08-20T14:39:15Z", "hours": 0.3869, "text": "Removed buggy pill_spin.js and used clunky animation pauses improved readme contents Restyled the site with different colour pallete adjusted layout for section 1 added drop shadow to the pfp image on hover removed a few dashed borders - replace some with solid border too!!"}, {"post_id": 50261, "ts": "2026-08-19T12:53:26Z", "hours": 3.6731, "text": "Made the whole site with 3 sections: The bit about me Language pill And projects This has taught me how to use api request with js and how coding in html+erb has hindered some of those abilities and remembering how to link scripts and stylesheets"}]...
|
Title: Personal Site by @Acidicts | Stardance
Dev Title: Personal Site by @Acidicts | Stardance
Devlogs:
Made the whole site with 3 sections: The bit about me Language pill And projects This has taught me how to use api request with js and how coding in html+erb has hindered some of those abilities and remembering how to link scripts and stylesheets
Removed buggy pill_spin.js and used clunky animation pauses improved readme contents Restyled the site with different colour pallete adjusted layout for section 1 added drop shadow to the pfp image on hover removed a few dashed borders - replace some with solid border too!!
Ships:
Made the whole Site: in 3 sections seen as etc to 3 with footer and some api usage for live hackatime info...
|
1793440070.4656115
|
1793526470.4656115
|
|
49813
|
My own site! by @LAZX | Stardance
|
LAZX
|
1
|
0.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T12:32:51Z", "hours": 0.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I made a simple portfolio like page according to the guide as I completely vibecoded this I didn’t have any challenges. I have not checked the page properly as its just a simple page showing basic info.", "repo": "https://github.com/Sakshyam-Xtha/personal-site", "demo": "https://sakshyam-xtha.github.io/personal-site/"}]...
|
[{"post_id": 50178, "ts": "202 [{"post_id": 50178, "ts": "2026-08-19T09:23:14Z", "hours": 0.2508, "text": "Finished the basic structure of the page."}]...
|
Title: My own site! by @LAZX | Stardance
Devlogs: Title: My own site! by @LAZX | Stardance
Devlogs:
Finished the basic structure of the page.
Ships:
I made a simple portfolio like page according to the guide as I completely vibecoded this I didn’t have any challenges. I have not checked the page properly as its just a simple page showing basic info....
|
1793465459.6492813
|
1793551859.6492813
|
|
48784
|
Bhavya’s Portfolio by @bhavyavashisth | Stardance
|
bhavyavashisth
|
1
|
7.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T12:23:18Z", "hours": 7.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "I have made a personal portfolio website using HTML,CSS and JS.The home page has the profile pic in gif formate,about me ,email,blogs and skills.Links of my github and x. There were some parts which were challenging: 1.The dark mode pfp: My first pass used a css filter to turn the gif pfp into a red silhouette,which worked terrible for my pfp like it makes the pfp straight up disappeared against the dark background. 2.Locking the colock to IST: First version used the viewer’s local timezone,so it read BST in London>fixed with Int.DateTimeFormate and timeZone:“Asia/Kolkata” Things of which i am proud of the most is how much good website i made with the most minimalism type work like it will only took 20 to 30sec to the viewer to scroll through the website and know about me if he is reading the blogs then story is different and the other things is color palette i chose for the website which goes perfectly with my pfp and looks good in both dark made and light mode in which light mode ain’t looking like a simple white layer rather then its something which is unique in my eyes. Things should people know so they can explore my site: Theme toggle is the small sun/moon button in the top-right. The ist clock on the home page is locked to Asia/Kolkata so it shows the right time no matter where you open it from.Updates every 30s", "repo": "https://github.com/bhavyavashisth/bahvyaportfolio", "demo": "https://bahvyaportfolio.vercel.app/"}]...
|
[{"post_id": 50238, "ts": "202 [{"post_id": 50238, "ts": "2026-08-19T11:55:02Z", "hours": 2.7031, "text": "devlog -#2 So i have to remove some of the things which i added last time as upcoming features: CVE Hobbies I removed them as no one cares what your hobbies and to be honest i tried and wrote the entire code but the results ain’t satisfying so i just deleted the code as i left in the website what i like , i really like to write blogs and kinda find it enjoyable to tell how did i and why did i kinda like it so i have added another blog “Reverse engineering a small CTF binary in C”. You can read the blog on the website if you find the title interesting. As of now the website is complete and i will try my best not make anymore changes “Minimalism is best imo”"}, {"post_id": 50076, "ts": "2026-08-19T04:07:41Z", "hours": 4.6636, "text": "devlog -#1 Ok so i am making my portfolio for 4 times now.This is one of the final struck so i don’t forgot what i did and why and everyone who is reading this can yell at me about what i got wrong lol. The goal: I am fed up from building a big ahh website therefor this time and one last time i want to make a minimalist website telling about me , profile pic, name,one line what i do a short paragraph, and a location pin with live clock that’s it. The kind of page you read in 20 sec and walks away knowing who i am. The stack html+css+ javascript The dark mode ate my pfp: ok so originally ideas was to have a character gif to turn into a red tinted shadow in dark mode. Looked good on paper but it was one of the bad idea which didn’t turned out to be as i thought so i need to swap the pfp with a simple gif of a my fav. character Reze from chainsaw men. The cropping of pfp: The gif i used was was a portrait jpg so the character’s face got chopped Fix: I just changed the focus of the pfp from full torso to face only which turned out to be good. The ist clock: This was a 10 line function and i still managed to mess it up, First version uses the local viewer’s local time zone so if anyone opened in london it would’ve showed the BST not IST. Fixed the clock to Asia/Kolkata timezone regarless of where visitor is.The clock updates every 30 sec, which is enough that no one notices the drift and cheap enough for not waste cpu. Hover animation: The page barely has any motion by default. I added a few small ones. Nav links get a red underline that grows from the left Skill cark lifts 2 px and turn their border into red when hover 3.The theme button rotates 15 degree when you point at it The pfp zooms 4% on hover, clicked the circle so it doesn’t leak The red glow bulb: There’s a fixed radial gradient in the top left corner that gives the page soft “Designed website” feel. What is left to do: need to write about my hobbies Add another blog as of now only wrote one fully and some coming soon blogs CVE Thats it the most time consuming thing was to write about me , blog and you can’t forget the clock Thats it, as of now will write about my hobbies , make a simple cve"}]...
|
Title: Bhavya’s Portfolio by @bhavyavashisth | Sta Title: Bhavya’s Portfolio by @bhavyavashisth | Stardance
Devlogs:
devlog -#1 Ok so i am making my portfolio for 4 times now.This is one of the final struck so i don’t forgot what i did and why and everyone who is reading this can yell at me about what i got wrong lol. The goal: I am fed up from building a big ahh website therefor this time and one last time i want to make a minimalist website telling about me , profile pic, name,one line what i do a short paragraph, and a location pin with live clock that’s it. The kind of page you read in 20 sec and walks away knowing who i am. The stack html+css+ javascript The dark mode ate my pfp: ok so originally ideas was to have a character gif to turn into a red tinted shadow in dark mode. Looked good on paper but it was one of the bad idea which didn’t turned out to be as i thought so i need to swap the pfp with a simple gif of a my fav. character Reze from chainsaw men. The cropping of pfp: The gif i used was was a portrait jpg so the character’s face got chopped Fix: I just changed the focus of the pfp from full torso to face only which turned out to be good. The ist clock: This was a 10 line function and i still managed to mess it up, First version uses the local viewer’s local time zone so if anyone opened in london it would’ve showed the BST not IST. Fixed the clock to Asia/Kolkata timezone regarless of where visitor is.The clock updates every 30 sec, which is enough that no one notices the drift and cheap enough for not waste cpu. Hover animation: The page barely has any motion by default. I added a few small ones. Nav links get a red underline that grows from the left Skill cark lifts 2 px and turn their border into red when hover 3.The theme button rotates 15 degree when you point at it The pfp zooms 4% on hover, clicked the circle so it doesn’t leak The red glow bulb: There’s a fixed radial gradient in the top left corner that gives the page soft “Designed website” feel. What is left to do: need to write about my hobbies Add another blog as of now only wrote one fully and some coming soon blogs CVE Thats it the most time consuming thing was to write about me , blog and you can’t forget the clock Thats it, as of now will write about my hobbies , make a simple cve
devlog -#2 So i have to remove some of the things which i added last time as upcoming features: CVE Hobbies I removed them as no one cares what your hobbies and to be honest i tried and wrote the entire code but the results ain’t satisfying so i just deleted the code as i left in the website what i like , i really like to write blogs and kinda find it enjoyable to tell how did i and why did i kinda like it so i have added another blog “Reverse engineering a small CTF binary in C”. You can read the blog on the website if you find the title interesting. As of now the website is complete and i will try my best not make anymore changes “Minimalism is best imo”
Ships:
I have made a personal portfolio website using HTML,CSS and JS.The home page has the profile pic in gif formate,about me ,email,blogs and skills.Links of my github and x. There were some parts which were challenging: 1.The dark mode pfp: My first pass used a css filter to turn the gif pfp into a red silhouette,which worked terrible for my pfp like it makes the pfp straight up disappeared against the dark background. 2.Locking the colock to IST: First version used the viewer’s local timezone,so it read BST in London>fixed with Int.DateTimeFormate and timeZone:“Asia/Kolkata” Things of which i am proud of the most is how much good website i made with the most minimalism type work like it will only took 20 to 30sec to the viewer to scroll through the website and know about me if he is reading the blogs then story is different and the other things is color palette i chose for the website which goes perfectly with my pfp and looks good in both dark made and light mode in which light mode ain’t looking like a simple white layer rather then its something which is unique in my eyes. Things should people know so they can explore my site: Theme toggle is the small sun/moon button in the top-right. The ist clock on the home page is locked to Asia/Kolkata so it shows the right time no matter where you open it from.Updates every 30s...
|
1792878461.0687482
|
1793742461.0687482
|
|
51253
|
Aditya's Portfolio by @adityajain1225 | Starda Aditya's Portfolio by @adityajain1225 | Stardance...
|
adityajain1225
|
1
|
1.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T11:43:45Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "This is my personal website I made it using TypeScript + HTML and bits and pieces of custom CSS :) This includes whatever I am passionate for and a wide list of the projects I do. Hope you will like it :)", "repo": "https://github.com/AdityaJainDXB/Aditya-portfolio", "demo": "https://aditya-portfolio-ten-eta.vercel.app"}]...
|
[{"post_id": 50142, "ts": "202 [{"post_id": 50142, "ts": "2026-08-19T07:43:45Z", "hours": 1.0994, "text": "Completed the website using NodeJS style"}]...
|
Title: Aditya's Portfolio by @adityajain1225 | Title: Aditya's Portfolio by @adityajain1225 | Stardance
Devlogs:
Completed the website using NodeJS style
Ships:
This is my personal website I made it using TypeScript + HTML and bits and pieces of custom CSS :) This includes whatever I am passionate for and a wide list of the projects I do. Hope you will like it :)...
|
1793480149.7117844
|
1793566549.7117844
|
|
50050
|
Tickets to The Moon by @Lokesh_Krishnan | Stardanc Tickets to The Moon by @Lokesh_Krishnan | Stardance...
|
Lokesh_Krishnan
|
1
|
2.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T10:59:10Z", "hours": 2.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "Ticket To Fly is a space travel booking app built entirely in a single index.html file. You can pick destinations like Mars or the Moon, customize your mission, and get a downloadable boarding pass with a working QR code. The hardest part was managing state, live pricing math, and saving past passes to localStorage using pure JavaScript with zero frameworks. Getting html2canvas to export clean PNGs was pretty frustrating to debug too. I’m super happy with how the UI turned out—especially the background starfield and the pass archive system (used Gemini & Perplexity to help brainstorm and troubleshoot parts of it). How to test: Pick “One-Way” or click outer planets to find easter eggs. Switch currency between USD and ★STARDUST at checkout. Finish a booking to download your pass, then click ARCHIVE in the header to view saved tickets.", "repo": "https://github.com/creationssptml8/Ticket-To-Fly", "demo": "https://effortless-conkies-ca4164.netlify.app/"}]...
|
[{"post_id": 50204, "ts": "202 [{"post_id": 50204, "ts": "2026-08-19T10:44:50Z", "hours": 0.4864, "text": "Website add-ons and Dynamic Pricing have been completed!"}, {"post_id": 49521, "ts": "2026-08-18T12:11:57Z", "hours": 0.4711, "text": "Website FINSIHED! Test it out!!"}, {"post_id": 48676, "ts": "2026-08-17T05:37:05Z", "hours": 0.9353, "text": "Finished the Ticket booking interface!"}]...
|
Title: Tickets to The Moon by @Lokesh_Krishnan | S Title: Tickets to The Moon by @Lokesh_Krishnan | Stardance
Devlogs:
Finished the Ticket booking interface!
Website FINSIHED! Test it out!!
Website add-ons and Dynamic Pricing have been completed!
Ships:
Ticket To Fly is a space travel booking app built entirely in a single index.html file. You can pick destinations like Mars or the Moon, customize your mission, and get a downloadable boarding pass with a working QR code. The hardest part was managing state, live pricing math, and saving past passes to localStorage using pure JavaScript with zero frameworks. Getting html2canvas to export clean PNGs was pretty frustrating to debug too. I’m super happy with how the UI turned out—especially the background starfield and the pass archive system (used Gemini & Perplexity to help brainstorm and troubleshoot parts of it). How to test: Pick “One-Way” or click outer planets to find easter eggs. Switch currency between USD and ★STARDUST at checkout. Finish a booking to download your pass, then click ARCHIVE in the header to view saved tickets....
|
1793490532.8607225
|
1793576932.8607225
|
|
50997
|
RISC-V Emulator by @jayanth_raveendra | Stardance
|
jayanth_raveendra
|
1
|
7.0
|
16
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T10:49:38Z", "hours": 7.0, "multiplier": 4.46, "payout": 16, "blessing": "neutral", "text": "A bare-metal, 32-bit RISC-V CPU emulator written entirely in Rust from scratch. It features a simulated 1MB memory array, a full RV32I base integer decoding pipeline, and Memory-Mapped I/O (MMIO) that intercepts writes to a specific memory address to simulate a physical terminal screen. Instead of just running standard assembly, I built a proxy-kernel into the emulator to intercept RISC-V ecall instructions (Syscall 64 for write and Syscall 93 for exit). Alongside a custom Makefile and linker script setup, this allows the emulator to natively execute freestanding, bare-metal C programs compiled via a Linux RISC-V cross-compiler toolchain.", "repo": "https://github.com/jayanth0725/risc_v_emu", "demo": "https://github.com/jayanth0725/risc-v-emu/releases/tag/v1.0.0"}]...
|
[{"post_id": 50203, "ts": "202 [{"post_id": 50203, "ts": "2026-08-19T10:41:20Z", "hours": 6.9356, "text": "Spent the almost 7 hours building a bare-metal, 32-bit RISC-V CPU emulator entirely in Rust from scratch! I finished the core RV32I instruction decoding and set up a simulated 1MB memory array, but getting actual C code to run on it was a massive challenge. I ran into two brutal bugs: The GCC cross-compiler was secretly compressing my instructions into 16-bit chunks, causing my 32-bit fetch() cycle to mash instructions together into unknown opcodes. I had to configure -march=rv32i to force strict 32-bit mode. My CPU tried to jump to an address 4.2 billion bytes out of bounds because of a 64-bit sign-extension trap in Rust! After fixing those, I wrote a tiny bootloader routine to set the Stack Pointer (x2), built a proxy-kernel to handle ecall system traps, and set up Memory-Mapped I/O (MMIO). The emulator can now successfully run a freestanding C program, print to a fake terminal screen, and exit cleanly!"}]...
|
Title: RISC-V Emulator by @jayanth_raveendra | Sta Title: RISC-V Emulator by @jayanth_raveendra | Stardance
Devlogs:
Spent the almost 7 hours building a bare-metal, 32-bit RISC-V CPU emulator entirely in Rust from scratch! I finished the core RV32I instruction decoding and set up a simulated 1MB memory array, but getting actual C code to run on it was a massive challenge. I ran into two brutal bugs: The GCC cross-compiler was secretly compressing my instructions into 16-bit chunks, causing my 32-bit fetch() cycle to mash instructions together into unknown opcodes. I had to configure -march=rv32i to force strict 32-bit mode. My CPU tried to jump to an address 4.2 billion bytes out of bounds because of a 64-bit sign-extension trap in Rust! After fixing those, I wrote a tiny bootloader routine to set the Stack Pointer (x2), built a proxy-kernel to handle ecall system traps, and set up Memory-Mapped I/O (MMIO). The emulator can now successfully run a freestanding C program, print to a fake terminal screen, and exit cleanly!
Ships:
A bare-metal, 32-bit RISC-V CPU emulator written entirely in Rust from scratch. It features a simulated 1MB memory array, a full RV32I base integer decoding pipeline, and Memory-Mapped I/O (MMIO) that intercepts writes to a specific memory address to simulate a physical terminal screen. Instead of just running standard assembly, I built a proxy-kernel into the emulator to intercept RISC-V ecall instructions (Syscall 64 for write and Syscall 93 for exit). Alongside a custom Makefile and linker script setup, this allows the emulator to natively execute freestanding, bare-metal C programs compiled via a Linux RISC-V cross-compiler toolchain....
|
1793440099.4879596
|
1793526499.4879596
|
|
51334
|
Personal-website by @harshitkataria20 | Stardance
|
harshitkataria20
|
1
|
1.0
|
0
|
[{"ship_num": 1, "ts": "2026-0 [{"ship_num": 1, "ts": "2026-08-19T10:35:22Z", "hours": 1.0, "multiplier": null, "payout": 0, "blessing": "neutral", "text": "i made my first personal website , it was little hard as i didnt knew much of js but i managed to pull it off", "repo": "https://github.com/hshit2010/Personal-website", "demo": "https://aboutharshit.info/Personal-website/"}]...
|
[{"post_id": 50201, "ts": "202 [{"post_id": 50201, "ts": "2026-08-19T10:33:31Z", "hours": 0.6567, "text": "My first personal website!! hope you all will like it:)"}]...
|
Title: Personal-website by @harshitkataria20 | Sta Title: Personal-website by @harshitkataria20 | Stardance
Devlogs:
My first personal website!! hope you all will like it:)
Ships:
i made my first personal website , it was little hard as i didnt knew much of js but i managed to pull it off...
|
1793487364.9732792
|
1793573764.9732792
|