Repository index
1016 AI, ML and Web3 projects · page 8 of 11
- ULTRACONTEXT FOR AI AGENTS
Ultracontext — open source context infrastructure for AI agents. Automatically captures and shares agent context everywhere.
- ARTEMIS AI COMPANION DESKTOP
ARTEMIS is an interactive AI companion with a live 2D model that generates images, answers questions and holds a conversation via GPT‑Sovits. Integrated with ComfyUI for customizing appearance and behavior.
- AGENT SHOPPING FOR MAXIMUM BENEFIT
Filtmall-Shopping-Skill — AI agent skill for finding identical products, checking actual price and automatic purchase. It provides checkout, order placement, delivery, and after-sales support.
- OPEN HARNESS PLATFORM FOR AGENT
OpenHarness — open-source platform for (agents). It lets you run Claude Code, Codex and any coding agents on your machines, training them to work in CAD, PCB, robotics and games, while the Harness hardware keeps them on your desk.
- СОЛО СКИЛЛЗ: AI АГЕНТЫ
solo-skills — набор продуктивности для 1‑person бизнеса: автоматизировано 49 процессов, включая 26 готовых навыков AI‑агентов со скриптами выполнения. Включает готовые скрипты запуска.
- PSTACK-CLAUDE AGENT CODE
pstack-claude moves Cursor primitives into ready-to-use workflows for Claude Code, Codex, Gemini and Prime Agent. The project is written in TypeScript, providing strict typing that lets you switch agents without changing code.
- KOOBOO JS TOOL FOR WEBSITES
Kooboo — JavaScript framework for CMS, websites, applications and ecommerce. AI agents generate code and logic from natural language; the built‑in module and CLI/API let you build fast and deploy projects.
- AI FOR NOVELS AND DRAMA
DSH plugin creates novels and short-form scripts with AI agents. Written in Python, uses Oh Story and Drama Skills.
- READ WITH AI: SMART READER
read-aware — an e-reader with a self-evolving AI agent that remembers your progress across all books and suggests contextual notes. Plugins extend format support, cloud integrations and agent features, and a single TypeScript codebase enables cross-platform operation.
- OPENGROK: MODEL MANAGEMENT
Run any model in Grok Bot with one command. Model selection UI and provider map with evidence‑based base. Update‑proof doctor guards against breakages on updates.
- SKILLS: AGENTS.MD TEMPLATE FOR AI
The project offers agent skills and an AGENTS.md template to isolate work in worktrees, build a service layer, validate with evidence, and release using proof‑before/after and Greptile review. Works with Claude Code, Cursor, and Codex.
- EASEL: AI AGENT FOR SOCIAL MEDIA
Easel — an open‑source AI agent that searches for trends, generates content and publishes it on Xiaohongshu, Douyin, Zhihu, BiliBili and other platforms. It learns from results, showing what really works.
- GENJUTSU CREATIVE CODE FOR CLAUDE
Genjutsu — a set of creative skills for Claude: animations, 3D graphics, design systems and motion principles. It lets you quickly create interactive visuals and prototypes in Python, integrating with Claude Code.
- ECOLOGIT: ENERGY ASSESSMENT
EcoLogits measures energy consumption and carbon footprint when using generative AI models via API. It collects CO₂ emission metrics and offers recommendations to reduce the environmental impact of LLM requests.
- PINLOOP-CLI: JOB SEARCH FOR AGENTS
pinloop-cli searches and scores over 10 000 jobs from 4 000+ companies, updating data hourly. The tool runs in the terminal and is suited for AI agents such as Claude Code.
- ACTUAL LIST OF LVLM HALLUCINATIONS
Curated list of recent works on hallucinations in large vision‑language models. Contains papers, datasets and tools for detecting and mitigating hallucinations. Regularly updated, includes over 300 stars on GitHub.
- MEMRA ARCHIVE CUDA INFERENCE
Rust + CUDA inference engine for RTX PRO 6000 Blackwell and RTX 5090. Serves safetensors and GGUF via OpenAI‑compatible API with per‑device tuned defaults and speculative decode byte‑identical to plain decode. Hosted at inference.tiyuvta.ai.
- CATALOG OF LLM TOOLS
llms-tools is a curated list of over 200 tools and projects for working with large language models. The project helps quickly find the right solutions for AI chatbots, data science and other tasks, providing links and brief descriptions.
- CREATE AGENT LIKE NEXT.JS
Library simplifies creating LangGraph agents like Next.js apps. Full TypeScript compatibility and TSX syntax.
- QA‑PILOT CHAT ASSISTANT FOR CODE
QA‑Pilot — an interactive chat that uses online or local LLMs for quick understanding and navigation of any GitHub repository. The project helps find functions, explain code, and generate tests using natural language.
- ROUTEVSCODE DYNAMIC AI SWITCHER
routeVSCODE — Zero‑reload switcher of AI models and local proxy for VSCode Copilot Chat via 9Router. Allows instantly switching GPT‑4, Claude, DeepSeek without reloading the editor, preserving the chat.
- WHISPER-WEBSITE: AUDIO TO TEXT
Simple web app on Python and FastAPI using OpenAI Whisper to transcribe audio into subtitles. Allows uploading files, selecting language and downloading SRT/VTT.
- PROTECTION AGAINST OVERDEFENSE OF AGENTS
HERO‑Anti‑OverDefense — a paste‑in contract that blocks four over‑defense patterns of agents. Works with Claude Code, Codex, Antigravity, Cursor, Copilot, Windsurf and Gemini CLI. Reduces false triggers and increases reliability of AI agents.
- OBSIDIAN TARS: AI TEXT BY TAGS
The Obsidian plugin generates text from tags using models DeepSeek, Claude, OpenAI, Gemini, Ollama and other LLM providers. It works entirely in TypeScript and integrates directly into notes.
- AI SEARCH OPENAI
A TypeScript sample app demonstrates the Retrieval Augmented Generation (RAG) pattern in Azure. It uses Azure AI Search to find data and Azure OpenAI with LangChain to generate ChatGPT‑style answers.
- ALFRED + OPENAI: SMART CHATBOT
Alfred 5 workflow to chat with GPT via OpenAI API. Generates and edits images quickly with DALL·E, converts speech to text and synthesizes voice 🔊
- XRAG: EVALUATION OF RAG MODULES
XRAG evaluates core components of RAG systems: indexing, retrieval, and generation. Using standard datasets it compares different module implementations, revealing their strengths and weaknesses.
- LOCAL-FIRST INBOX FOR AI AGENTS
pizza-bot is a local-first inbox for long-running AI agents on DeepAgents and LangGraph. Supports human-in-the-loop and runs via Electron. 319 stars on GitHub.
- FINANCE GURU: SELF-HOSTED FAMILY OFFICE
Self-hosted family-office engine: typed Python calculators, a private SQLite ledger, and agents for Claude Code and Codex. Agents propose solutions, typed code executes them.
- BLD — PAINLESS JAVA BUILD
BLD — a pure Java utility that builds projects without configuration files. Just specify the source folder, and it will automatically find .java files, compile them, and package them into a JAR. Supports incremental builds and dependency management via Maven repository.
- PMETAL: THE ACCEL LLM ON APPLE SILICON
PMetal — Rust framework for high‑performance LLM inference on Apple Silicon. Supports local inference, LoRA/QLoRA fine‑tuning, serving, quantization, and MLX/Metal acceleration and swift.
- GAC: SMART AI COMMIT GENERATOR
gac is a Python tool that uses an LLM to generate meaningful commit messages based on changed code. It supports many languages and integrates easily into any workflow via GitHub Actions or a local CLI.
- NIM SERVER: LSP BASED ON NIMSUGGEST
langserver — the Nim language server implementation based on nimsuggest. It provides the code completion, go-to definition, the diagnostics and code formatting, while supporting both the LSP, MCP and the protocols.
- DEXТ: MODERN FULL‑STACK FOR DELPHI
Dext is a modern full‑stack framework for Delphi that combines asynchronous tasks, channels and dependency injection. It provides an Entity‑Framework‑like data API and ready‑to‑use components for web and desktop apps.
- APOLLO GRAPHQL MCP SERVER
Rust MCP server for Apollo GraphQL. Gives LLMs access to schema, operations, and registry via standard interface. Speeds up AI agent development for data tasks.
- SKILLD: SKILLS FOR AI AGENTS
Curated agent skills by humans. Search, run, install, and keep them current from one CLI. Works with Claude Code, Codex, Cursor and Gemini CLI.
- PERSISTENT MEMORY FOR AGENTS
funes stores AI agent session history in a searchable LanceDB database. Lets you find context from past dialogues and code in milliseconds. Written in Rust, integrates with Claude Code and Codex.
- TTY‑TABLE: TABLES IN TERMINAL
TTY‑TABLE is a cross‑platform library for creating ASCII tables in terminals on Windows, Linux, and macOS. It also works in the browser console and, via the MCP server, lets you quickly add ready‑made tables to README and other files.
- DSH-HUB-CLI PLUGIN MANAGEMENT
Open-source TypeScript CLI for DSH Plugin Hub: generates schemas, resolves dependencies, and provides DSH agent tools. Speeds up creation, updating, and publishing of plugins from the terminal and integrates with AI agents.
- AGENTAO LOCAL-FIRST AI RUNTIME
Agentao — a local-first AI agent runtime for Python with access control, MCP protocol, memory and audit replay. Run it as a library in your app, as a CLI tool, or as an ACP server; everything works offline, no data leaves your machine.
- HEADCOUNT AGENTS CLOUD DEV TOOLKIT SYSTEM
Headcount — a firm agent organization for Claude Code, structured as a company: 15+ departments, 125+ skills, each installable independently. Enables swift connection of needed tools via the online agent marketplace.
- GOGOGOT: AI AGENT ON GO
GoGogot — lightweight self‑hosted AI agent in Go that you can run locally and use instead of OpenClaw. It understands natural instructions, executes bash commands, works with the file system, and connects to local LLMs via an OpenAI‑compatible API.
- SUPRESEO CLAUDE SKILLS
11 Claude skills for SEO: page audits, link building, article writing, E‑E‑E‑T audit and semantic gap analysis. Methodology based on Koray Tuğberk, Kyle Roof and Lily Ray approaches, supplemented with anti‑AI‑slop rules at generation time and tested at InhouseSEO.
- MCP FISCAL SERVER FOR BRAZIL
MCP server for fiscal operations: processes CNPJ, NF-e, NFS-e, CT-e, SPED, eSocial, Simples Nacional and accounts for Reforma 2026. Provides 44 tools, works without registration and uses offline data. Implemented in Python.
- CONTROL UNITY EDITOR WITH AI
UnityMCP lets you control the Unity Editor via MCP over HTTP straight from the terminal or an AI agent. The isuzu-unity-cli command needs no Node, .NET or other runtimes — everything runs inside the editor itself.
- AI AGENT FOR SUPPORT QUALITY EVALUATION
chat-quality-agent is a Go-based AI agent for evaluating customer service quality. It analyzes chats from Facebook Messenger and Zalo to identify weak spots in support. The project has already gained 296 stars.
- DOBERMAN-CORE — PROTECTION FOR AI AGENTS
Doberman-Core intercepts every input, output, and tool call during AI-agent operation, blocking unsafe actions. It works as middleware, checking authorization and applying guardrails without modifying agent code.
- AWESOME-TYPESAFE OFFICIAL RESOURCES FOR AI
A curated list of official resources and community projects for TypeSafe, System One models, and Jev. Includes links to documentation, code examples, and tools for working with LLM and structured output.
- BUILD-CLI: YOUR TERMINAL FOR MICROSOFT BUILD
Utility Build-CLI integrates Microsoft Build into the terminal. Works with GitHub Copilot CLI, allowing you to launch talks and sessions directly from the command line. Supports TypeScript and has open source code.
- VIDEO GPT-PLUS: ADVANCED VIDEO UNDERSTANDING
Official repository of the VideoGPT+ paper: combines image and video encoders for improved video understanding. Enables extracting semantic features, generating descriptions, and answering video questions using GPT-4o.
- RIPWIRE: CODE MAP FOR AI
ripwire — zero‑dependency C++23 CLI + MCP server that in seconds builds a deterministic ranked map of repo code, computes blast radius and suggests the set of tests affected by changes. It also shows quality deltas of complexity and coverage between commits, saving ~5% of tokens compared to grep‑reading.
- OPEN PROTOCOL FOR AI AGENT CONTRACTS
OpenWorkProof provides an open protocol for creating and verifying AI‑agent work contracts. It offers cryptographic authorization, evidence collection for execution, and formal acceptance of results in multi‑agent systems.
- PROMPTS FOR GPT-4V AND DALL-E3
Awesome-Multimodal-Prompts — selected collection of prompts for GPT-4V and DALL-E3 that unlock full multimodal capabilities. Use them to improve image understanding with GPT-4V and boost image generation quality with DALL-E3.
- BOCOEL: BAYESIAN OPTIMIZATION FOR LLM
bocoel — a Bayesian Optimization framework for quick LLM evaluation: 10× faster, just a few lines of code. It replaces costly benchmarks with precise search over model hyperparameter space.
- OPALINE: AI SESSION ANALYTICS
Opaline — CLI for analyzing Claude Code and Codex sessions. Shows token stats, costs, and usage patterns in seconds. 288 GitHub stars.
- DOMAINSTACK.IO SMART DOMAIN SEARCH
DomainStack.io — all‑in‑one domain name intelligence service: checks availability, DNS data, AI‑based suggestions, and access control. Everything in one TypeScript API, ready for Hacktoberfest.
- USEAGENT — AI COWORKER FOR TEAM
useagent — AI coworker for your team: agent provides cloud environment, access to tools and context, returns websites, presentations, spreadsheets, reports and pull requests. Runs Claude Code, Codex and OpenCode using your subscription.
- CHENGOS VISUAL AI WORKFLOW
ChengOS — visual system for AI data processing and agent workflow construction. The repository contains deployment scripts, infrastructure configs, and runtime resources for production and development. Enables quick launch of AI agents with Claude Code integration.
- CODEX PETS FROM PHOTOS
Creates high-fidelity Codex companions from two or more photos using local open-weight neural network weights. No OpenAI API key needed — everything works offline.
- LLM EVALUATION THE LATEST MATH OLYMPIADS
matharena evaluates LLMs on fresh math olympiad problems. It loads the problems, runs models, via API, and, measures the solution accuracy in JSON.
- MONITORING AI AGENTS IN REAL TIME
Local observability map for AI coding agents: shows plans of Claude Code, Codex and Cursor, their tool‑calls, file changes and progress in real time. Lets you instantly see what the agent is doing, without extra logs.
- MULTI-PROMPT FOR LLM ON DESKTOP
llm-god — a desktop app built with Electron/TypeScript that lets you send simultaneous requests to ChatGPT, Gemini, Claude and DeepSeek. Compare answers in real time, save sessions and export results with one click.
- FLARE: GRAPH-BASED AGENTIC IDE
Flare is an IDE where code is represented as an interactive agent graph. You can visually link LLM‑controlled modules, set dependencies, and run complex workflows directly from the editor. It supports TypeScript and is extensible via plugins.
- AI REDTEAM AGENTS PLATFORM
AI red‑team platform: autonomous LLM agents perform a full pentest in an isolated Kali container and generate a report. LangGraph plan/act engine, models via LiteLLM, output PDF/JSON/SARIF, interface FastAPI + React.
- QCODE: ISLAND FOR AI CODE
QCode — interactive island where you learn AI coding, building real projects together with AI agents. The platform runs on DeepSeek Harness and is written in TypeScript.
- RHINOAI: SMART AGENT FOR RHINO
RhinoAI adds AI capabilities to the Rhino modeling environment via MCP‑agents in C#. The project enables geometry generation, automates routine tasks, and integrates LLM prompts directly into the interface.
- MNEMORY: MEMORY FOR AI AGENTS
Self-hosted, secure, feature-rich memory for AI agents and assistants. Extracts facts intelligently, removes duplicates and stores artifacts with full content.
- FAST LAUNCH QWEN3.8 ON AMD
The project provides scripts for instant deployment of Qwen3.8-Flash-Next on AMD Strix Halo (gfx1151) GPU via HIP. All dependencies are built automatically, and inference runs with maximum throughput.
- JLLM: GPU‑ACCELERATED LLAMA3.JAVA
Project jllm implements Llama3.java inference on the GPU using the TornadoVM while staying pure Java. Supports the GGUF format and models DeepSeek‑R1 and Granite. It has 276 stars on the GitHub.
- NEURALGRAFFITI: LIVE LLM OUTPUT CONTROL
NeuralGraffiti is a tool for live transformation of foundation model outputs directly at the neural network level. Works in real time without retraining.
- INTLLINODE: UNIFIED GATEWAY TO AI
IntelliNode provides a unified prompt layer for accessing the newest AI models: ChatGPT, LLaMA, Deepseek, Diffusion and others via Hugging Face. Built-in performance evaluation lets you compare latency, cost and answer quality in real time.
- AI AGENT COORDINATION PROTOCOL
foremerge — open coordination protocol for coding agents over Git. Catches intent conflicts before code conflicts appear. Written in Rust, works via CLI and MCP.
- MEMORY + CONTEXT FOR AI AGENTS
CLI is a tool for orchestrating memory and context for coding agents. It stores interaction history, injects relevant code and suggestions in real time, boosting generation accuracy.
- SMS: SIMPLE MBTILES SERVER IN PYTHON
SMS — lightweight Python server for serving vector tiles from MBTiles over HTTP. Supports tile requests via TMS/XYZ standard, delivers GPX tracks as overlay and is MCP-compatible for AI‑agent integration.
- RESUME-JOB MATCHER: AI CANDIDATE FILTER
TypeScript CLI for AI resume screening: hard constraints filter out unfit candidates before scoring, each score cites the resume span, and an embedding pre‑filter makes screening large pools cheap. Supports Claude, GPT or OpenRouter.
- 🔍 LIGHTHOUSE: VIDEO MOMENT RETRIEVAL
User-friendly library for reproducible video moment retrieval and highlight detection. Also supports audio. Published at EMNLP2024 Demo, ICASSP 2025 and ICASSP 2026.
- BLOOM AI TUTOR FOR EVERYTHING
Bloom — an AI tutor that analyzes your learning style and delivers the next lesson tailored to you. It implements Bloom 2‑Sigma as a Claude Code skill and offers a self‑hostable web tutor focused on Chinese language and Socratic dialogue.
- BRIDGE QQ AND AI
qq-bridge connects the QQ bot based on SnowLuma OneBot v11 with DeepSeek Harness agents, providing social simulation, safe MCP tools, and slang learning. It allows easy extension of the chatbot's functionality via AI agents.
- OPENAI PYTHON WITH X402 PAYMENT
The openai-python library is a drop‑in replacement for the official OpenAI SDK on Python, adding transparent payment via the x402 protocol. It supports agents, Ethereum and Solana, as well as standard LLM requests.
- COTAL: OPEN STANDARD FOR AGENTS
Cotal is an open standard for coordinating AI agents, defining a communication protocol and forming dynamic agent meshes. It enables skill discovery, composition, and management of agent workflows in TypeScript.
- AWESOME-OPEN-AI-SORA: RESOURCE COLLECTION OF SORA
The Awesome-Open-AI-Sora list gathers articles, videos, podcasts and news about the Sora model from OpenAI. Here you will find up-to-date materials on its capabilities, applications and latest updates.
- ALPACA TRADING AGENT
AlpacaTradingAgent — a multi-agent LLM framework for automated trading via Alpaca API. It supports market data analysis, trading signal generation, and order execution in real time. Runs on Python and easily integrates with any LLM models.
- OPEN BROWSER FOR AGENTS
open-browser-use — cross‑platform browser automation for AI agents. It offers real Chrome control via CLI and SDK with no vendor lock‑in, ready to use out of the box.
- PLUGIN GOALS FOR OPCODE
The plugin adds reliable goal workflows to OpenCode: it persists state, limits actions with safety limits, and provides agent tools. Completion occurs only upon confirmation of evidence of execution.
- MIMEO: TURN ANY EXPERT INTO AN AGENT
Create AGENT.md and SKILL.md from expert knowledge in your repository. Mimeo analyzes code, documentation, and practices, turning them into instructions for AI agents. Supports Claude Code and CLI.
- PI-DESKTOP: LOCAL AI CODER
Local AI coding agent: Electron + Rust core + OpenAI, Claude and local model support. Fast repository indexing via tree-sitter, plugins and i18n.
- PYSCRAPPY: ADAPTIVE WEB-SCRAPER
PyScrappy — an adaptive Python toolkit for web scraping with an MCP server for AI agents. Self‑healing selectors and TLS‑fingerprint stealth evade anti‑bot filters, while 24 built‑in scrapers deliver clean LLM‑ready data from any URL.
- OPEN STEPS FOR AI
Project open-steps provides a set of skills that turn a coding agent's output into plain‑language text: honest reports, clear verdicts, and actionable steps. All skills are released under the MIT license.
- OPENCLIP — FLOATING TEXT ON MAC
OpenClip — a fast open‑source tool for macOS that shows floating actions over selected text. It lets you transform text, run scripts and connect AI with no extra steps.
- CLAUDE SKILLS FOR FOUNDERS
Claude‑skills‑founder — a set of Claude Code skills that help startup founders create product briefs, conduct competitor analysis, shape pricing strategy, and prepare pitch decks and GTM plans straight from the terminal.
- LAUNCH YOUR OWN AGENT NETWORK
Huddol lets you create and manage your own agents, combining them into workflows. It supports agent runtime, AI workspace, and flexible workflow configuration via Python.
- BASE: AI OPERATING SYSTEM
BASE turns Claude Code from a one-shot tool into a workspace that remembers context, self-maintains, and never gets stale. This is Builder's Automated State Engine in Rust, ensuring persistent agent state across sessions.
- ALL SKILLS FOR AI AGENTS
awesome-skills — community of Apify agent skills for AI coding assistants. 252⭐, Python, includes ready‑made skills for actors, agents and MCP workflows.
- TRAINING LLM FOR SCIENCE
ScholarCopilot trains language models to create scientific texts with accurate citations. The project is based on COLM 2025 research and implemented in Python.
- SHIP STUDIO: DEVELOPMENT WITHOUT LIMITS
Ship Studio is a local-first development environment where you can write, test, and deploy code without leaving the app. It works with any CLI‑based coding agent such as Claude Code, Codex, or Cursor and supports TypeScript projects.
- PLUGINS CLAUDE CODE FOR RESEARCH-PLAN-IMPLEMENT
The ed3d-plugins repository has Claude Code plugins implementing the research-plan-implement workflow. It allows the plan generation automatically creating installation instructions & extending capabilities via Python-scripts.
- REDUCE AI TOKEN COSTS
Chisle cuts AI-agent token usage: terse prose, YAGNI-first code, and tool-output compression. Works with Claude Code, Pi, Cursor, Codex, Gemini and four more, zero deps, with full benchmarks.
- MERIDIAN: AUTOMATIC WORK JOURNAL
Meridian keeps an automatic work journal: it records all completed tasks and changes, daily creates a brief report and generates draft tickets in your tracker. It runs on Rust and integrates with AI‑agents (Claude, ChatGPT) for accurate summarization.
- SPEC‑DRIVEN AI WORKFLOW
Autonomous workflow for AI agents: reconnaissance → plan → build → validation → release with TDD, adversarial validation, and anti‑slop quality gates. Packaged as Kiro Power, written in Shell.
- CODEROCKET UI COMPONENT LIBRARY
Open-source component library for React and Vue written in TypeScript. It includes 48 ready‑made components, 26 UI blocks, design tokens, a CLI and MCP support. Allows visual UI customization via CodeRocket UI.