chessapi4j

Hi! I started this project last year and have been improving it, fixing errors and bugs along the way. This year, I picked it up again and made several improvements. It is a Java library that allows you to work with chess-related operations.

The library includes features like verifying legal positions and legal moves, working with PGN files, generating legal moves from a given position, and handling FEN strings.

Since chess is a game, many natural abstractions arise. Developing a relatively fast move generator was particularly challenging (I believe Java can't offer much more speed), while also ensuring it passed the standard unit tests for a move generator. The library includes many features, and perhaps I'm not doing a good job of describing them all.

I am just an amateur programmer, and I wrote this library because I wanted to create a server for an online chess website. As I started, I realized it was better to develop it as a separate project from the server, and then I realized it didn't have to be proprietary software, so I decided to share it, which led me to GitHub in the first place.

Astroboard the Umbraco dashboard

Repo: https://github.com/wpplumber/astroboard

Get insights of your contents, assets and members while using the friendly CMS.

No special blockers except Angular JS used by the CMS for services.

I'm a web programmer focused on Front-end with experience in developing Umbraco websites and packages also web apps using Tailwind CSS, SPA and PWA in Vue (+ Quasar Framework, Vuetify), AstroJS.

Professional aligned with agile methodology. 📆

fan_control (https://github.com/TarikTornes/fan_control)

Hey everyone I am an undergrad student from Luxembourg and I am new to the OpenSource world and to C and therefore decided to create a small project which is a daemon called fan_control.

It started when I wanted to build my own little DIY homeserver with a Raspberry Pi 4B, and decided to cool it with a small cooling system which uses a fan. It bothered me that the only option was to run it permanently on full speed, so I decided to build my own little project in order to fix my issue and improve my C knowledge by coding a small daemon which checks the temperature of the Raspi and adjusts the fan speed according to the temperature.

I am sharing it with you to hopefully get feedback and suggestion according to my style of coding, efficiency and everything about creating a OpenSource project, since I am quite new to this.

Feel free to give me feedback of any kind, wether it is a comment below or as an issue on my repo.
You can find the project here

PS: You can also give feedback to my other repos.

I built FishCode: a novel reader that looks like a code editor.

The idea started with a playful question: what if a novel looked like text data inside a SQL file?

Import TXT, EPUB, or ZIP files, and chapters become files in an editor-style sidebar. You can switch between SQL, JSON, YAML, Markdown, Python, C++, and TypeScript views, adjust the reading layout, and export a self-contained HTML file for offline reading.

Local imports stay in your browser. No signup required.

Project and screenshots:
https://github.com/zhao1411935907/fishcode-moyu

I’d love feedback on two things: which view feels most natural, and how comfortable the layout is for longer reading sessions.

Hi everyone!

I’m excited to share my first project here: Sales Report Generator.

https://github.com/dev-celia/sales-report-generator

This is a small project I created while learning programming and working on my GitHub portfolio. The idea was to build something that could generate a sales report from data and, at the same time, practice writing cleaner and more organized code.

One of the main challenges for me was understanding how to structure the project and getting used to Git and GitHub along the way. I’m still learning, so I know there is plenty of room for improvement!

I’d really appreciate any constructive feedback or suggestions on how I could improve the project, especially the code structure and README.

I’m looking forward to building more projects and seeing how much I can improve over time!

Hi! I'm sharing a small, early-stage project: TaskClear Prompt by mirdobrr-hash.

Repository: https://github.com/mirdobrr-hash/universal-assistant-prompt

The idea came from wanting an AI assistant to understand the actual task instead of blindly following fragments of a request. TaskClear is a Markdown prompt that asks the assistant to clarify important ambiguity, preserve requirements, perform authorized work, verify results and state limitations.

The repository includes a full Russian prompt, a shorter Russian version, an English translation, and five example tasks with observable evaluation criteria. You can paste the prompt into a chat and then provide your task. It does not install tools or add capabilities to the model.

Current limitations: there are no published model comparisons yet, and a redistribution license has not been selected. The examples are test inputs, not evidence of effectiveness. The project and this introduction were prepared with AI assistance.

My goal is to develop this into a useful, honestly evaluated resource. The main challenge now is measuring whether the extra instructions actually help rather than just making answers longer.

Feedback I'd appreciate: which task would you use to compare the full prompt, the short version and no prompt at all? Reproducible failures and suggestions for clearer instructions would be especially useful. Thanks!

style-contract: a Claude Code skill that keeps AI images in one consistent style.

Ask GPT Image or Nano Banana for ten images and you get ten styles. This skill writes a short "style contract" for your project (exact hex palette, finish, composition, subject rules, mood), builds every prompt from it, and checks each render before it ships. It includes 5 presets (riso zine, park poster, bauhaus, ukiyo-e, y2k chrome).

https://github.com/tk1475/style-contract

Feedback very welcome, and a ⭐ if it's useful!

My new asymetric game for 2 players. Not balanced atm.
https://michalstankiewicz4-cell.github.io/PatientZero/

Hi, I'm William (@starchlorddotcom). I'm sharing Lag Doctor, a free, MIT-licensed Paper Minecraft plugin developed with substantial AI assistance. This introduction was also prepared with AI assistance.

The problem it aims to help with is: after changing one thing on a lagging server, did the tick measurements actually improve? It records tick-spike statistics, keeps recent slowdown incidents, optionally scans already-loaded entity chunks, and compares a baseline with fresh measurement windows. Reports export as local TXT and HTML; nothing is uploaded. It complements spark profiling and does not automatically fix lag or prove which plugin caused it.

There are separate builds for Paper 26.3 / Java 25+ and Paper 1.21.11 / Java 21+. The repository now includes browsable source, tests, installation instructions and actual reports from a labelled synthetic demo. Recorded validation includes 18 automated tests and local Paper smoke checks; production-scale overhead and real-world diagnostic accuracy still need independent testing.

My goal is to turn this early project into something server owners find useful and build a public portfolio along the way. Feedback I'd value: is the installation clear, and does the report help you choose a next investigation step? If you run Paper, try the matching JAR on a test copy, wait about a minute of active ticks, then run /ld report and /ld export.

Downloads and quick start · Testing feedback thread. Thanks for taking a look.

S

OpenKey: Building an Open-Source WebHID Driver for Mechanical Keyboards ⌨️

Hi GitHub Community! 👋

I've been working on OpenKey, an open-source project exploring browser-based customization and hardware control for mechanical keyboards using WebHID.

🔗 Live demo: https://openkey.bakul.tech/
💻 Source code: https://github.com/BakulBd/OpenKey

Why I started this project

Many mechanical keyboards offer customization features through manufacturer-specific software. I wanted to explore whether an open-source, browser-based alternative could make those features more accessible, especially for Linux users and people using keyboards without convenient cross-platform tools.

What I'm working on

• Investigating WebHID communication with mechanical keyboards
• Exploring support for the AJAZZ AK820 Max Plus
• Building a foundation for compatibility with additional keyboard models
• Learning about USB HID reports, device permissions, and hardware protocols

Looking for community feedback

I'd love to connect with developers experienced in WebHID, USB HID, Linux, and hardware reverse engineering.

If you own an AJAZZ or another mechanical keyboard, I'd be interested in hearing which model you use and what customization features you'd like an open-source tool to support.

This is an ongoing project, and device compatibility is still being investigated. I'd appreciate your feedback, suggestions, and contributions!

Thanks for reading! 🚀

Maze Signal — learning weighted pathfinding with Python, C and a browser demo

Hi everyone! I'm returning to sharing my programming projects publicly and would like to make them useful for people learning algorithms. Maze Signal is a small, AI-assisted open-source project comparing Dijkstra and A* on weighted grids.

• Repository: https://github.com/mohammmad357/maze-signal
• No-install demo: https://mohammmad357.github.io/maze-signal/
• Current release: https://github.com/mohammmad357/maze-signal/releases/tag/v1.2.0

The problem that inspired it: the route with the fewest steps need not have the lowest cost. The demo shows route cost and expanded nodes; those numbers are not a Python/C runtime benchmark. A recent addition exports the edited map into the same text format used by the Python and C command-line implementations. Keeping the three implementations consistent has been the main challenge.

I'm looking for 5–10 independent first-use testers. Please open the demo before reading the README and tell me:

1. Where did you get stuck or feel confused?
2. What was useful?
3. What was missing?

A one-off, optional invitation to fellow participants whose shared projects overlap with this work:

• Python/C and command-line tools: @therealbenflynn @herlonventura @dev-celia @TarikTornes
• Teaching, algorithms and explaining code: @kevin-lozada-santos @lunalobos
• Browser interfaces and games: @zhao1411935907 @Dotoryman @michalstankiewicz4-cell
• Interpreting measured results: @starchlorddotcom

If this fits your interests and you have a couple of minutes, a brief first impression here or a reproducible issue in the repository would help. No obligation to respond. The project and this invitation were prepared with AI assistance; independent usability feedback is still pending. My goal is clearer explanations and a maintainable learning tool. Thanks for reading!

Creator Rail — looking for independent demo testers

Hi everyone! I'm sharing Creator Rail, an MIT-licensed Python/Web prototype for a Persian/English creator support page, developed with AI assistance. I wanted to explore a clearer, self-hostable support flow with direct-to-creator payments as the eventual goal.

Repository and setup: https://github.com/mohammmad357/creator-rail

The current release is a local synthetic demo: no real funds, wallet connection, API keys or wallet secrets are needed. It includes invoice/QR previews, exact satoshi amounts, payment-state scenarios, CSV export and backup/restore. Keeping partial payments, manual status changes and duplicate events from being mistaken for confirmed payments was a key challenge. The BTCPay adapter has mock-based tests; live processor integration is still pending. Mainnet is disabled.

I'm looking for three independent first-use testers. Please follow the README's demo setup without my help, switch languages and try a few synthetic invoice scenarios. Then tell me:

1. Where did you get stuck or feel confused?
2. What was useful, and what was missing?
3. Which OS/browser did you use, and can you reproduce any problem?

A reply here or an issue in the repository would help. Please don't send money or share secrets/private data. This is an experimental prototype, not a live donation service. No obligation to respond; honest criticism is welcome. Thanks!

I'm sharing DraftScope, an Apache-2.0 research library for reproducible NVIDIA LLM inference experiments: https://github.com/6bjjjj/draftscope

The project grew out of experiments on how numerical uncertainty in a speculative verification block triggers an expensive dense output-head fallback. Its experimental extension conditions that decision on the prefix that actually determines greedy acceptance. It uses existing vLLM, DFlash2 and Marlin components, with CUDA conditional graphs for the branch.

Practical features include a Transformers baseline, capability-based FP16/BF16 selection, pinned model revisions, input hashes, incremental result saving and a bounded CLI worker. English/Arabic documentation and the original experiment records are included.

One useful lesson from the A100 experiments: a component improvement does not automatically translate into a large model-level gain. The captured head study reduced fallback calls from 33 to 9 and head wall time by 18.95%; the later whole-model V8 result was 0.922% lower request latency. Different numerical guards were used, and the independent quality audit remains inconclusive. The refactored package still needs GPU integration testing.

This is an AI-assisted research project, with attribution and limitations in the repository. My goal is to make the experiments inspectable and invite reproducible feedback, rather than claim a new state-of-the-art decoder.

Feedback that would help: is the separation between CPU-tested interfaces and historical GPU evidence clear, and what minimal numerical or compatibility test would you add before enabling the optional head on another NVIDIA architecture? There's a GPU-report issue template for anyone who wants to contribute a result.

Super Street Fighter Turbo Remake — a free fan project for Windows

Hi everyone! I’m edhard-remakes, a solo gaming enthusiast and a newcomer to GitHub. I enjoy revisiting classic games, and I’d like to share my Street Fighter II fan project with the community.

The game uses IKEMEN GO, with a custom interface and a roster of 16 characters and their stages. I’ve just published the first experimental Windows build, v0.1.0. It is still a work in progress.

This is a free, non-commercial fan project. My goal is to share what I make, learn from others and encourage fellow fans to create things too. I’m also using AI tools to support the creative and experimental process. Credit goes to the IKEMEN GO team and the community authors whose characters, stages, sounds and other resources form part of the project. This is an unofficial project, with no affiliation with Capcom.

• Project: Super Street Fighter Turbo Remake
• Windows download: Experimental release v0.1.0 — the game is the .exe under Assets (about 1.23 GB).
• Feedback and bugs: Open an issue. I added Portuguese templates for bug reports and opinions or suggestions.

I’d love to hear what you think of the visuals, controls and combat feel. If you try it and find a bug, please tell me what happened and how to reproduce it. You don’t need to be a programmer to help — playtesting and honest feedback are very welcome!

Thanks for taking a look, and I’m looking forward to meeting other fighting game fans and creators here.

Hi everyone! I'd like to share dsh-mcp-bridge, my first open-source project that I'm actively maintaining and sharing with others.

I built it to make it easier to delegate tasks from Claude Code, Codex and other compatible MCP clients to DeepSeek Harness. Repeated calls can leave a lot of sessions to manage, and I wanted model, workspace and permission settings to remain visible and controllable in the Harness UI.

The plugin adds an in-app control panel, MCP task status and results, reusable sessions, and automatic session archiving. A running DeepSeek Harness host is required. Version 0.2.4 is available as a preview; model calls, toolbar configuration inheritance and read-only permissions have been tested.

Repository: https://github.com/FishCityBee/dsh-mcp-bridge
Release: https://github.com/FishCityBee/dsh-mcp-bridge/releases/tag/v0.2.4

If you already use Harness with an MCP client, I'd appreciate feedback on setup friction and which tasks you find useful to delegate. The repository includes English and Chinese documentation.

Tenchu Shadow Reborn — an experimental fan project for Windows

Hi everyone! I’m edhard-remakes, a newcomer to GitHub and a fan of classic games. I don’t know how to code; AI tools are what made it possible for me to start experimenting with game creation, and I want to be open about that.

My project revisits Tenchu: Stealth Assassins, exploring HD textures and changes to the camera and combat. Current experiments include a free camera, target lock, guard and timed parry, stamina and posture break, heavy attacks and stealth rolling. It is still a work in progress, and the full campaign has not been validated.

I do this for fun, with no commercial goals. I’ve enjoyed many community creations over the years, and I’d like to give something back by sharing my progress and learning from other fans. Credit belongs to the original game’s creators and the authors of the tools and resources involved. This is an unofficial fan project.

• Project and development overview: Tenchu Shadow Reborn
• Feedback, ideas and bugs: Open an issue

I’d love to meet other Tenchu fans and hear your thoughts. You don’t need programming experience to participate — opinions, ideas and honest feedback are welcome!

ntroducing MoneyMate My Android Finance App

I built MoneyMate, an Android app designed to make everyday money management easier. It helps users track their income and expenses in one place.

I developed it using Kotlin and Jetpack Compose, with help from AI tools.

🔗 GitHub Repository: https://github.com/manzoorshoro/MoneyMate

I'm sharing my project with the community and would appreciate your feedback and suggestions!