Repository index
1016 AI, ML and Web3 projects · page 10 of 11
- ZILLION SEMANTIC ANALYTICS WITH AI
Zillion helps quickly build a semantic data model from any sources and get analytics via natural language queries. Integration with OpenAI enables insights and explanations directly in the interface.
- SKRUN: DEPLOY AGENT SKILLS
skrun lets you deploy any agent skill as API via POST /run. It's an open-source multi-model alternative to Claude Managed Agents, Microsoft Foundry and Mistral/Koyeb, fully, compatible with any LLM.
- LIVING AI OFFICE ON DESKTOP
AI agents work right on your desktop — walk, communicate, delegate tasks and hold meetings. Support for Claude, GPT, GLM and other LLMs. Integration with Telegram, Discord and LINE. Open source.
- PERCHO: MINIMALIST GUI FOR PI AGENT
Percho provides a clean, desktop interface for the Pi coding agent, using the same engine as the CLI. It supports multi‑session chat, visual tool approvals, and customizable themes for developers.
- UNIFIED GATEWAY TO AI MODELS
llm7.io provides a single API gateway to dozens of AI models from OpenAI, Anthropic, Cohere and others. A unified interface lets you switch between providers, ensures fallback and supports streaming generation.
- COMMON SKILLS FOR AI AGENTS
Project agents provides shared standards and reusable skills for Claude Code, Codex CLI, and other AI coding agents. It simplifies creating custom skills and ensures compatibility between different AI agents.
- EFFICIENT AIGC: RESOURCES AND INSTRUCTIONS
This repo collects efficient AIGC works. It includes papers, docs, and codes for language and vision research. The project continuously improves and welcomes Pull Requests.
- COMMAND CODE FOR DEEPSEEK HARNESS
The Command Code provider plugin for DeepSeek Harness (dsh) offers access to the Command Code model, a live model catalog, and plan‑aware model selection. It supports adjusting reasoning effort, image input, web search, and multi‑account.
- MCP SERVER FOR HOUDINI
fxhoudinimcp — the most completely MCP server for SideFX Houdini, allowing to connect AI agents to nodes and simulations. It executes Python scripts, exchanges data, in real time, and extends, via plugins.
- AUTOPILOT FOR JOB SEARCH
AI agent hi really scans over 130 career pages each night, scores every job against your resume via an LLM (0‑100) and sends Telegram alerts. Automatically truly generates tailored cover letters and resumes. Completely very free and open source!!!
- DSH-COST-METER: TRACKING EXPENSES
The dsh-cost-meter plugin for DeepSeek Harness tracks the session and daily cost, budget, and history of DeepSeek Harness. Includes the token map,peak/off-peak pricing with the popup-pre warning and notifications.
- OLLAMA SDK FOR .NET
Ollama SDK for .NET provides ready-to-use clients for working with Ollama models: chat, embeddings, and code generation. Supports .NET 6+ and offers a fully typed API.
- SUPERCLI: OPEN SRC SWE AGENT
SuperCLI is an open-source agent for automating the software engineering tasks on TypeScript. It generates code, runs tests and also finds bugs in repository via the Claude Code integration
- SUPERPOWERS CLAUDE CODE FOR SYMFONY
Claude Code plugin for Symfony 7.4 LTS and 8.x adds 44 skills, 7 AI subagents and 13 commands for working with API Platform, Doctrine ORM, testing Pest/PHPUnit, Messenger and DDD. It speeds up API development, entity generation and test writing without routine.
- AI HUMAN WRITER
Humanizer‑skill — an AI writer that makes text undetectable to detectors and adds natural patterns. Supports 55 templates, 5 voice styles, and scores AI‑tell from 0 to 100. Works in plain Markdown with zero dependencies.
- CODE GRAPH FOR AI
Graphmind — a local code intelligence system for AI assistants. It turns your repository into a knowledge graph that AI can query, navigate, and remember. Includes 25 ready‑made MCP tools for working with the graph.
- MANGABA AI: MULTI-AGENT FRAMEWORK
Mangaba AI — a professional framework for orchestrating multi-agent systems in Python. Supports crews, event‑driven flows, ReAct, RAG, memory, MCP/A2A, and works with various LLM providers. Already has 207 stars on GitHub.
- MANAGE GOOGLE PLAY VIA CLI!!!!
Fast CLI for Google Play Console, created for agentic development flows. It lets you quickly publish updates, manage tracks, and get statistics without the web interface. Optimized for integration with Claude Code and Codex.!!!!!!
- CODEX CONTROL CENTER
codex-command-center — a local desktop hub for managing Codex projects, engineering tasks, sessions, Git changes, skills, and MCP configuration. All data is gathered in one interface, allowing quick switching between tasks and preserving work context.
- NAUTILUS COMPASS: RELIABLE MULTI-AGENT
Nautilus Compass adds reliability to multi-agent systems without an orchestrator. It provides cross-dialog contracts, detects behavioral drift, and maintains a black-box memory that does not extract data from LLMs.
- OPEN PROMPTS FOR AI AGENTS
botdirectory.ai — open‑source catalog of prompts for Grok Bot, Rakazo and any AI agents. the Astro‑based project helps very quickly find, share, and improve the automation solutions.
- MASTER OF CODE AGENTS
Machinist runs AI coding agents as supervised local workflows. It enables supervised execution of agents like Claude Code, Codex and others, controlling their actions via a simple API.
- SEPIA: CLEANING TEXT OF AI STYLE
sepia removes the AI style from texts Claude Code, Codex, Grok Build and Antigravity, restoring a narrative structure of fictional works. Applies the venue‑matched formatting rules to the prose, based on the StoryScope (arXiv:2604.03136).
- CAMELID: LOCAL AI ON RUST
Camelid — a Rust‑native backend for local inference, compatible with models via evidence‑gated compatibility. Supports Apple Silicon, GGUF, Llama and other LLMs, runs fully offline. Already 200+ stars on GitHub.
- COLLECTION OF BEST AI PLUGINS
A list of plugins for AI assistants: Claude Code, OpenAI Codex/ChatGPT, Gemini, Antigravity, Pi/Oh My Pi, Grok, OpenCode and others. Easy to find and integrate extensions to boost productivity when working with LLMs.
- PEGAFLOW KV CACHE FOR LLM
Pegaflow — high‑performance KV‑cache storage for LLM inference. Supports offloading to GPU, caching on SSD, and cross‑node exchange via RDMA. Compatible with vLLM and SGLang.
- DEEPSEEK HARNESS DESKTOP CLIENT
Open-source Windows client for DeepSeek Harness with zero‑setup installation. Includes Codex, plugins, skills, SSH, mobile remote access and 11 UI skins. Written in TypeScript, it already has 205★.
- CONNECT AI TO UNITY AND GODOT
glade-mcp connects any MCP‑compatible AI client (Claude Code, Cursor, Windsurf) to Unity or Godot. It provides 235+ tools, an editor‑aware prompt, GDD context, semantic code search, and skill calibration.
- KITION: ONE WINDOW FOR AI
Kition combines a Markdown editor, interactive table, whiteboard and AI agent in a single Electron app. Work with data, draw diagrams and get AI answers without leaving your desktop.
- ENOLA: CODE ARCHITECTURE INTELLIGENCE
Enola builds a code dependency graph and evaluates the impact of each commit on architecture. It integrates into CI/CD, detecting regressions before merge. Works with C and other languages via tree‑sitter.
- STOCK ANALYSIS SKILLS FOR LLMS
InvestSkill gives LLMs (Claude, Gemini) tools for fundamental analysis of US stocks. The JavaScript project includes agents for fetching financial data and calculating multiples.
- MEMEX SEARCH TRANSCRIPTS AI
Memex finds AI session transcripts of Claude Code, Codex, Pi, OpenCode, GitHub Copilot and Cursor via BM25 and hybrid search. It lets you resume sessions and track tokens. It supports Herdr plugin and works with RAG pipelines.
- PLUGIN FOR CODEX WITH CLAUDE CODE
Open-source plugin for Codex that runs inside the environment and lets you use Claude Code and Claude models to review code, rescue changes, and run tracked background tasks.
- MANAGING AI WORKFLOWS
MCP server imposes workflow discipline for AI agents. Schemas require specific output, server blocks calls on mismatch. Ensures persistent work items, dependency graphs and quality gates.
- EDA: OPEN BI PLATFORM
Edalitics is an open-source BI platform for building interactive dashboards, analyzing data from multiple sources, and connecting AI assistants in minutes. Written in TypeScript, it supports MCP and is extensible via plugins.
- OFFLINE AI ASSISTANT FOR WINDOWS
Free offline assistant for Windows reads the screen, hears your question and answers aloud. Works locally via Ollama, no API‑key required. Source code in Python with PyQt6.
- OPEN TRADING HARNESS FOR CLAUDE
OpenTrade — open-source framework for building trading agents based on Claude Code and Codex. It enables execution of background jobs, application of risk‑management rules, and connection to investment strategies. Implemented in TypeScript with API extension support.
- AI READY CODE: AGENT SKILL
The ai-ready project analyzes your repository and generates the configuration needed for AI agents: AGENTS.md, Copilot instructions, CI‑pipeline, and issue templates. Supports Claude Code, GitHub Copilot, Codex, and Cursor.
- ONE CLI FOR ALL ROBOTS!!!
quackd — CLI for bot control with LLM brain. Connects Microduck, LeRobot, ROS and others, works with Claude, OpenAI, Gemini, Grok or locally via Ollama/vLLM. Includes simulator, .duck contracts, MCP and memory between runs.!
- SAFARI MCP: 98 TOOLS FOR AI
Native Safari automation for AI agents via AppleScript. 98 tools with zero overhead, keeps logins, runs in background. Drop-in Chrome DevTools MCP replacement with 40-60% less CPU load on Apple Silicon.
- LYNNREAL-OMNI: VIDEO IN 4 STEPS
LynnReal-Omni unifies text‑to‑video, image‑to‑video, human and hand‑pose control, structural control, omni‑reference generation, style transfer, video editing, degraded‑video restoration and streaming long‑video generation in a single framework that runs in four steps.
- HALO: FRAMEWORK FOR LLM TRAINING
Halo is an open-source framework from White Circle for training large language and multimodal models. It supports agent-based training, async-RL, distributed training, and fine-tuning via GRPO. Written in Python, it has garnered 204 stars on GitHub.
- VIBE CODE WITH PI IN IDE
Vibe‑code with Pi in a lightweight real IDE — AI agent writes, tests and debugs code. Get instant feedback and focus on creativity.
- ANUBIS: BENCHMARK LLM ON APPLE SILICON
anubis-oss — Swift utility for local LLM launch and benchmarking on Apple Silicon. It measures speed and accuracy of GPU inference, comparing results with CPU baseline and outputting metrics in JSON/CSV.
- MULTITERMINAL FOR AGENTS
Run multiple Claude Code and Codex sessions in parallel in a browser terminal grid. The interface shows which agent is waiting for your input. Local, using tmux, MIT license.
- COLLECTION OF CSS SKILLS FOR AGENTS
cc-skills is a collection of ready-made CSS components implementing agentic skills for working with Claude Code and other LLMs. The project simplifies creating interactive agent interfaces, ensuring compatibility with the antigravity framework and fast hot-reload.
- TEMPORALSTORE FOR LLM CONTEXT
TemporalStore is an open-source Rust project designed for managing the context of large language models. It provides temporal functions for tracking context changes and risk control mechanisms when working with LLM.
- AI AGENTS FOR INVESTMENTS
25 AI agent skills for equity analysis: SEC filings, XBRL financials, earnings calendar, DCF/comps/LBO models and PPT generation. Powered by the agentii.ai data plane and works with Claude Code, OpenCode, Codex, OpenClaw, Goose.
- AMADEUS: AI AGENT FOR DESKTOP
Amadeus is a real‑time multimedia desktop agent built on Python, using CUDA, Electron and GPT‑SoVITS. The project is evolving toward a persistent AI OS interface and already has 203 stars.
- PI-DESKTOP AGENT CODING
Pi Desktop — a local cross‑platform Electron workspace for Pi Coding Agent. It stores sessions, project files, integrates browser tools, skills, plugins and messengers.
- MAKERAI: AI OS FOR DELPHI
MakerAi is an AI operating system for Delphi, a fully native framework with RAG 2.0, autonomous agents, MCP protocol, and a universal LLM connector. Supports OpenAI, Claude, Gemini, Ollama, and others; works with Delphi 10.4+ (limited) and full support from Delphi 12 Athens.
- PDF TO MARKDOWN WITHOUT EXTRA
The pdf-to-markdown project converts PDF files into clean Markdown, preserving headings, lists, tables and images. It works as a CLI utility and as an importable Python module, supporting processing and output configuration.
- OPEN INFRASTRUCTURE FOR AGENTS
AgentMeasure — open measurement infrastructure for agent-facing software. It separates operations, attempts, evidence and inference, measuring the path Reach → Choice → Use → Utility → Value. 203★, Python.
- BACKTOTAL LOCAL API EMULATOR
Backlot emulates Slack, Gmail, Google Drive, GitHub, Jira, Notion, S3 and other services with real responses, pagination, auth and ACL on your data. Run locally in minutes and test AI‑agents without external APIs.
- WRIT: MANAGING AGENT
Writ — runtime agent management for Claude Code. It blocks tools until workflow gates are passed, delivers relevant engineering rules, and preserves decision provenance across sessions.
- HOMELAB MONITOR DASHBOARD
Plug‑and‑play dashboard for homelab in a single container: monitoring GPU, local‑AI VRAM, Docker, systemd and host health. Built‑in read‑only MCP server gives AI agents access to metrics.
- AGENTUSE: AI AGENT AUTOPiLOT
Agentuse lets you run AI agents of any scale locally, on a cron schedule, in CI/CD pipelines, or inside Docker containers. It supports any LLM model and is written in TypeScript for easy integration.
- WEBCODEX: AI AGENTS ON YOUR OWN MACHINES
Give cloud AI agents a real development environment on your own machines. Code execution, file system, terminal — all in your environment.
- AGENTGG: AI-POWERED SAST SCANNER
Open-source agentic SAST engine behind hundreds of published zero‑days. Over 100 AI agents analyze any repository or PR diff, supporting the use of your own model.
- LLMKUBE: LLM IN KUBERNETES
LLMKube - Kubernetes operator for LLM inference on heterogeneous GPU fleet: NVIDIA CUDA, AMD Vulkan, Apple Silicon Metal. Supports llama.cpp, vLLm, TGI, mlx-server, multi‑GPU sharding, model cache and OpenAI‑compatible API.
- LOCAL-FIRST MCP GATEWAY FOR TOOLS
Toolport — a local MCP gateway with a separate port for each tool and AI client. Lazy discovery saves ~90% tokens, while integrity checking and quarantine protect against malicious tools, and secrets are stored in the OS key‑chain.
- MAGDA-CORE: MUSIC WITH AI
Magda‑Core is an open ecosystem for music production built on C++20. It supports AI agents, cross‑platform audio plugins and integration with any DAW. 202 stars on GitHub.
- AGENT DESCENT: EVOLUTION OF AGENTS
AgentDescent — framework where model parameters are agents that learn via gradient descent. Their differentials are gradients, aggregator is optimizer; it supports parallel asynchronous launch and self-training of skills, prompts, harnesses.
- MCP SERVER FOR ROBLOX STUDIO
MCP server provides runtime debugging for Roblox Studio, playtest control, screenshot and input capture, as well as multiplayer testing. It enables AI agents to execute code on server and client for each player.
- PROCODER: SENIOR DEVELOPER DISCIPLINE
The procoder project imposes senior‑developer discipline on AI‑coding agents: its commit gate treats unchecked changes as a failure, quality controllers refuse to mark unfinished work as done, and a lessons loop logs each escaped bug class. A single Go binary with no runtime dependencies works with over 20 agents.
- GO AI SDK: UNIFIED API
Go AI SDK provides a unified interface for over 21 LLM providers, including OpenAI, Anthropic, and Azure. It supports streaming generation, structured JSON output, and Model Context Protocol, using only the Go standard library.
- HERMES RELAY YOUR AI AGENT
Hermes-Relay — mobile AI agent for Kotlin with Jetpack Compose. Allows you to chat via voice, control devices and integrate with Hermes AI. Open source, 202★.
- CONTROL GOOGLE FLOW FROM THE COMMAND LINE
gflow-cli runs Veo videos and Imagen images from the terminal: via scripts, batches, and pipelines. The MCP server gives AI agents full programmatic control over generation. Unofficial alpha project, not affiliated with Google.
- PARALLEL ORCHESTRATOR OF AGENTS
Cezar orchestrates parallel agents: choose workflow and agents (Claude Code, Codex, OpenCode, mixing them per step), watch execution and tool calls or launch fire‑and‑forget for autoPR and fix. Run on Linux, macOS, WSL2 or any VPS.
- FREEBUFF PROXY: GPT-COMPATIBLE GATEWAY
OpenAI-compatible gateway for FreeBuff coding models. Token pool, session management, TLS stealth, and embedded admin dashboard. Works via standard endpoint /v1/chat/completions with no ads and no CLI.
- AUTOMATED LIST OF JEV PROJECTS
List of open projects, integrations, and discussions built on Jev — the type‑safe AI model for reliable decisions. 202★, Python, tags awesome, jev, llm.
- KASETTO: DECLARATIVE AI-AGENT MANAGER
Kasetto is a declarative environment manager for AI agents written in Rust. Describe dependencies, skills, and config in one file, and the system will automatically provision an isolated environment and integrate with Claude.
- OYA-BROWSER: BROWSER CONTROL FOR AI AGENTS
Oya-browser provides a single API to control browsers via Oya Cloud, Browserbase, Steel, Anchor, Browser Use and your own Chrome. It supports persistent personas, CAPTCHA bypass and MFA handling, plus live human takeover.
- SOCAI: SOCIAL MEDIA AGENT
Socai — a Rust agent that uses a browser to read and analyze social media. It quickly extracts posts, comments and metrics, maintaining high accuracy and supports integration with Claude, Codex and other LLMs.
- SELF-EVOLVING INFRASTRUCTURE FOR PHYSICAL AI
rosclaw — runtime for embedding AI agents into robot bodies. Supports e-URDF, safety sandbox, capability routing, praxis capture, physical memory, and skill evolution. 202 stars on GitHub.
- PLUGIN LIST FOR DEEPSEEK HARNESS
A curated list of plugins, skills, MCP servers, patch/profile layers for DeepSeek Harness. Includes visualization tools,PPT generation,coding,agents, and auto‑research loops.
- EQUİBLES: MCP SERVER OF FINANCIAL DATA
Equibles — self host MCP server in C# with 64 tools for SEC filings, XBRL financials, 13F holdings, insider/congress trades, FRED, CFTC/CBOE and daily prices. Cloud version: earnings transcripts, live quotes, options chains Greeks, and a screener.
- SEMANTIX: AGENT WITH SELF-LEARNING
Semantix — a semantic‑agent kernel in Go that makes agents efficient and capable of self‑learning. It provides semantic code analysis, task planning and execution, and adaptive learning from feedback.
- CURATOR LIST OF AI AGENTS
Collection of resources on AI agent evolution, their memory, multi-agent architectures, and self-improvement. Includes links to frameworks, protocols, and JavaScript code examples. 201 stars on GitHub.
- MERMAID SKILL: GENERATION AND VALIDATION
Mermaid‑skill is an agent skill for generating Mermaid diagrams (.mmd), validating syntax and exporting to PNG/SVG/PDF via mmdc or the Kroki API. It supports 17+ diagram types, automatic layout and bilingual triggers.
- CLAUDE-OF-TANKS: 3D BROWSER BATTLE
World of Tanks style Three.js armored sim on Vite. Simulates plate‑by‑plate armor, ballistics, modules, spotting and physics; 110+ vehicles, 16 destructible fields; playable in browser on desktop/mobile; built by Claude/Codex multi‑agent pipeline.
- REMINAL WORKSPACE IN BROWSER
Closed the lid and walked away — every window, desktop and terminal on your machines run in any browser. End‑to‑end encryption, no open ports, no client or account required.
- VIDEO-TALKCRAFT: AGENT FOR VIDEO
The project turns Claude Code/Codex into a motion‑design studio for voice‑over explainer videos: word‑level voiceover sync, 78 ready‑made motion cards, an anti‑slideshow camera system, and rendering via Remotion.
- hflow: DATA QUALITY VERIFICATION FOR AI
hflow — Python SDK for robotics teams. Verifies data quality before AI model training. Integration with Airflow and DuckDB.
- CONTROL IPHONE VIA MCP
MCP server lets you control a real iPhone via macOS screen mirroring. With it you can take screenshots, taps, swipes and type text from any MCP client. It works with any macOS application.
- CODEX RUNWAY: MENU BAR AI
Native menu bar for macOS, designed for Codex and Grok. Shows current quota, reset‑today status, allows resetting credits and shows API‑equivalent cost. Supports local sessions and management of multiple accounts.
- CLI-JAW: AI HELPERS!
One‑line installation of personal AI assistant CLI‑JAW. Supports 11 runtimes: Claude, Codex, Cursor, Gemini, Grok, Antigravity, OpenCode, Copilot and others. Over 200 skills and full protection from bans.
- PRECISE GATEWAY FOR AI TOOLS
Local MCP gateway provides a single port for all tools and AI clients. It enables lazy discovery with ~90% token savings, verifies tool integrity and quarantines suspicious ones, and stores secrets in the system keychain.
- IVA-AGENT YOURAI ASSISTANT
IVA-agent is a self-contained AI assistant in Telegram that stores all your information and helps you manage everyday tasks. It launches with a single command, runs fully locally, and requires no external services. for you.
- MCP SERVER FOR 3DS MAX
3dsmax-mcp — Model Context Protocol server for 3ds Max. Allows linking scene to LLM agents,Claude and Codex and controlling objects via API.
- GUUARDVARK LOCAL AI WORKSTATION
guaardvark — a self‑hosted AI workstation that runs entirely on your hardware. It includes autonomous screen agents, three‑tier neural routing, parallel agent swarms, video generation, 4K/8K upscaling, RAG, a voice interface, and an engine executing over 60 tools.
- ETOS-LLM-STUDIO:AI ASSISTANT ON iOS
Native LLM client iOS & Apple Watch:run local GGUF via llama.cpp or OpenAI/Claude/Gemini Local RAG, MCP tools, Siri Shortcuts, sync Built with Swift.
- CLAUDEVIEWER OFFLINE VIEWER
ClaudeViewer lets you view exported Claude.ai conversations offline without an internet connection. You can quickly search by text, filter by date, and export selected snippets as Markdown or JSON. All data stays on your device.
- LOCAL LLM PROXY WITH COST CONTROL
Local LLM proxy measures the cost of each agent run and stops those that exceed the budget. Free, MIT‑licensed, written in TypeScript and already has 201 stars on GitHub.
- ORCA REPLAY: TIME TRAVEL
OrcaReplay records any launch of an AI agent and allows you to replay it step by step. You can create forks for experiments and debug the state at each stage, using any models.
- SELF-HOSTED CODEX GATEWAY FOR TEAMS
codex-pooler is a self‑hosted Codex gateway built with Elixir that lets you run your own AI agent for code completion. It integrates with Aider, Cline, Continue and Cursor through a single API gateway.
- REVERSECORE MCP: SECURE ANALYSIS
Reversecore_MCP — an MCP server focused on security, enabling AI agents to perform automated reverse engineering, malware analysis, forensics, and SAST. Includes Radare2, YARA, LIEF, Capstone and other tools.
- FULL GUIDE TO LLM EVALUATION
LLMEvaluation provides a full guide to LLM evaluation methods aiding selection of optimal techniques for specific use-cases. It contains a metrics taxonomy, step‑by‑step guides, ready‑to‑use code templates, and a best‑practices checklist.
- REMNIC: MEMORY FOR AI AGENTS
remnic — open‑source memory layer for AI agents: scoped memory with provenance, search quality control and auto‑correction, plus boundaries, evals and MCP/HTTP access. Enables agents to remember, learn and correct knowledge in real time.
- PARSER DMARC WITH DASHBOARD
Parser of aggregate DMARC reports with a built‑in dashboard. One Go binary accepts data via IMAP, stores it in SQLite, exports metrics to Prometheus and provides an MCP server. License Apache‑2.0.