| OpenKB | ★ 4.6K | RAG | Python | llm · knowledge-base · rag · retrieval · wiki · document-ingestion · vectorless-retrieval · pageindex · multimodal · cli · agent-skills · obsidian · rest-api · web-ui · python · litellm | OpenKB is a Python CLI that compiles raw documents (PDF, Office, HTML, CSV, URLs) into a persistent interlinked markdown wiki using LLMs and PageIndex vectorless tree retrieval. It offers query/chat with citations, a Skill Factory for agent skills, a web UI and REST API, all via LiteLLM providers. | View Project → |
| mcp-context-forge | ★ 4.5K | MCP | Python | mcp · gateway · registry · proxy · a2a · federation · rest-to-mcp · grpc · observability · opentelemetry · plugins · rate-limiting · auth · fastapi · kubernetes · tool-calling | ContextForge is IBM's MCP gateway, registry and proxy that federates MCP servers, A2A agents and REST/gRPC APIs into a single endpoint with auth, rate limiting, plugin support and OpenTelemetry tracing. Deployable via PyPI or Docker for centralized agent/tool governance. | View Project → |
| Octop | ★ 4.5K | AI Agents | Python | self-hosted · multi-agent · multi-user · long-term-memory · rag · mcp · acp · local-first · privacy · im-integrations · plugins · browser-automation · cron-scheduling · personas · knowledge-base · cli | Octop is a self-hosted, multi-user AI assistant platform with multi-agent support, running as one Python process (FastAPI) that serves a web dashboard, CLI, IM channels, and cron. Includes RAG knowledge base, portable memory, MCP connector gateway, ACP IDE/coding-agent delegation, browser automation, and plugins. | View Project → |
| awesome-harness-engineering | ★ 4.5K | AI Agents | Python | agent harnesses · AI agents · context engineering · agent memory · orchestration · MCP · agent evaluation · permissions · observability · templates | A curated collection of resources, patterns, and templates for engineering reliable AI agent harnesses. Covers planning, context, memory, tools, permissions, orchestration, evaluation, and observability. | View Project → |
| gptme | ★ 4.4K | AI Agents | Python | cli · terminal-agent · autonomous-agents · coding-agent · mcp · computer-use · local-llm · multi-provider · tool-use · plugins · skills · memory · rag · tui · webui · self-hosted | gptme is a provider-agnostic terminal AI agent that writes code, runs shell/Python, edits files, browses the web and sees images via tool use. Ships CLI/TUI, web UI, server API, MCP/ACP support, plugins and skills; runs fully local with llama.cpp. | View Project → |
| ruby_llm | ★ 4.4K | AI Frameworks | Ruby | ruby · rails · llm-framework · agents · tool-calling · multimodal · embeddings · rag · structured-output · streaming · image-generation · audio · video · provider-agnostic · mcp · async | RubyLLM is a Ruby-native AI framework giving one consistent API for 18+ providers: chat, streaming, embeddings, RAG, tools, agents, structured output, images, audio, video, and OCR, with first-class Rails integration and cost tracking. | View Project → |
| skills | ★ 4.4K | Skills | JavaScript | agent-skills · claude-code · codex · mcp · skills-cli · orchestration · visual-plan · visual-recap · agent-native · prompt-pack · workflow · code-review · autonomy · multi-agent · plugin | BuilderIO/skills is a collection of small composable agent skills (Claude Code, Codex, Cursor, Copilot) for docs-first discipline, autonomous execution, cross-agent review, visual plans/recaps, and frontier-model orchestration. Installed via the @agent-native/skills CLI or a plain skill-folder copy. | View Project → |
| Unity-MCP | ★ 4.3K | MCP | C# | unity · mcp-server · game-development · gamedev · ai-agents · cli · csharp · claude-code · cursor · copilot · editor-automation · runtime-ai · tool-generation · skills · llm | Unity MCP server plus CLI and skill generator that lets Claude Code, Cursor, Copilot or any MCP client drive the Unity Editor and runtime. Ships 70+ built-in tools, lets you expose any C# method as a tool, and supports both editor automation and in-game LLM use. | View Project → |
| mcp-server-cloudflare | ★ 4.3K | MCP | TypeScript | mcp · mcp-server · model-context-protocol · cloudflare · remote-mcp · streamable-http · workers · monorepo · oauth · observability · ai-gateway · autorag · browser-rendering · dns-analytics · developer-tools · typescript | Monorepo of Cloudflare's remote MCP servers exposing domain-specific, typed tools for Workers bindings, observability, docs search, Browser Run, AI Gateway, AutoRAG, DNS analytics and more over stateless Streamable HTTP at /mcp. Connect any MCP client (Cursor, Claude, OpenAI Responses API) to automate Cloudflare tasks | View Project → |
| archestra | ★ 4.3K | MCP | TypeScript | MCP gateway · MCP registry · LLM gateway · AI agents · agent orchestration · A2A · guardrails · RAG · OAuth · RBAC · observability · Kubernetes | An enterprise AI platform combining an MCP gateway and private registry with an LLM gateway, agent runtime, orchestration, guardrails, and observability. Supports Kubernetes deployments and integrations with common LLM providers and developer clients. | View Project → |
| nono | ★ 4.3K | AI Agents | Rust | agent-sandbox · zero-trust · agent-security · mcp-security · credential-proxy · code-execution · least-privilege · l7-filtering · policy-engine · llm-security · sandbox · supply-chain-security · sigstore · rust · ffi-bindings · audit | nono is a zero-setup, zero-latency Rust sandbox for running AI coding agents and their delegated tools under least-privilege policy. It isolates filesystem, network and credentials per agent and per tool (including MCP servers), with no VM, daemon or container. | View Project → |
| SemIf-OpenJev | ★ 4.2K | AI Tools | Python | semantic decisions · direct logits · open models · model inference · probability calibration · decision scoring · benchmarking · WebGPU | Python tool for scoring runtime-defined semantic decisions with open models. It reads option probabilities directly from model logits, with calibration and direct, cached, and parallel scoring modes; includes PyTorch, MLX, llama.cpp, and browser demos. | View Project → |
| excel-mcp-server | ★ 4.2K | MCP | Python | excel · spreadsheets · workbooks · file-manipulation · stdio · streamable-http · formulas · charts · data-validation · VBA · Claude · Cursor | An MCP server that lets AI assistants create, read, and edit .xlsx and .xlsm workbooks without Microsoft Excel. Supports spreadsheet formatting, formulas, charts, data validation, and VBA inspection, with stdio and Streamable HTTP transports. | View Project → |
| awesome-harness-engineering | ★ 4.2K | AI Agents | — | agent harnesses · harness engineering · coding agents · context engineering · agent evaluation · observability · safe autonomy · multi-agent systems · benchmarks · curated resources | A curated collection of guides, benchmarks, specifications, and open-source projects for making AI agents more reliable. Covers context and memory, evaluation, observability, safe autonomy, coding-agent harnesses, and runtimes. | View Project → |
| KiroCrew | ★ 4.2K | AI Agents | Python | agent · ai-agents · persistent-workspace · local-ai · automation · devtools · llm · slack · discord · multi-channel · self-improving · scheduling · skills · mcp · desktop-app · cli | Kiro Crew is a local-first persistent AI agent workspace for development work. Chat via CLI, web dashboard, desktop app, or Slack/Discord; run unattended multi-step tasks, recurring jobs, and heartbeats that keep working across sessions. | View Project → |
| yichen-skills | ★ 4.2K | Skills | Python | agent-skills · claude-code · codex · skill-pack · wechat · x-twitter · asr · transcription · playwright · obsidian · markdown · local-first · privacy · content-archive · video-editing · automation | A Claude Code / Codex Agent Skills collection for creators: X Articles draft uploads, X post slicing to video, WeChat/WeCom local data capture, ASR transcription, ChatGPT web research, and agent memory vaults. Includes a bounded Codex x ChatGPT review loop with MCP. | View Project → |
| ragent | ★ 4.1K | RAG | Java | agentic-rag · rag · mcp · spring-ai · java · multi-channel-retrieval · intent-recognition · query-rewrite · session-memory · rerank · vector-search · knowledge-graph · react-agent · deep-thinking · langfuse · enterprise | Production-grade Java Agentic RAG platform covering document ingestion, multi-channel retrieval (vector/keyword/graph/web) with RRF and rerank, intent recognition, query rewrite, session memory, MCP tool calling and a ReAct agent engine. Ships a React admin console, tracing and enterprise reliability features. | View Project → |
| jev-chat-jarvis | ★ 4.1K | AI Agents | Kotlin | android · kotlin · accessibility-service · chat-assistant · llm · wechat · qq · feishu · ocr · ml-kit · overlay · intent-detection · reply-suggestion · privacy-first · openrouter · screen-reading | Android accessibility-based chat copilot that reads on-screen WeChat/QQ/X/Feishu conversations and uses configurable LLM endpoints to infer intent, score risk, and draft ranked reply candidates that it fills into the input box but never sends. Read-only, no hooks, local knowledge base and history. | View Project → |
| skill-recorder | ★ 4.1K | Skills | TypeScript | agent-skills · skill-generation · workflow-recording · desktop-automation · screen-capture · Copilot CLI · Electron · task-reconstruction · workflow-automation · privacy | An Electron desktop app that records on-screen work and uses GitHub Copilot CLI to reconstruct the task as an intent and ordered steps. Review the result and generate a reusable agent skill or automation. | View Project → |
| DeskcommCRM | ★ 4.1K | AI Agents | TypeScript | crm · whatsapp · ai-agents · self-hosted · multi-tenant · mcp · rag · sales-automation · chatbot · customer-support · lgpd · nextjs · supabase · waha · typescript · ecommerce | Self-hosted open-source CRM where native AI agents attend, qualify and sell over WhatsApp (WAHA). Multi-tenant, LGPD by design, MCP-ready with RAG, deploying via a one-command VPS installer. | View Project → |
| basic-memory | ★ 4.1K | MCP | Python | mcp · mcp-server · ai-memory · knowledge-management · markdown · knowledge-graph · semantic-search · local-first · obsidian · claude · codex · cursor · zettelkasten · privacy-first · python · embeddings | MCP server that turns plain Markdown notes into a bidirectional, local-first knowledge graph AI clients can read, write, and semantically search. Adds persistent memory to Claude, Codex, Cursor, or any MCP host, with optional Obsidian and cloud sync. | View Project → |
| chat-on-steroids | ★ 4.1K | MCP | TypeScript | mcp · chatgpt · chrome-extension · electron · local-first · multi-agent · computer-use · coding-agent · browser-control · desktop-automation · cross-platform · terminal · durable-workflows · developer-tools | Cross-platform Electron app plus Chrome extension that gives ChatGPT local MCP capabilities: file editing, terminals, tests, desktop and browser control. Adds durable multi-agent workers, Goal/Loop task tracking and Compact & Resume for long coding sessions. | View Project → |
| Hyper-Extract | ★ 4K | RAG | Python | knowledge-graph · hypergraph · information-extraction · rag · graphrag · lightrag · semantic-search · embeddings · llm · cli · langchain · faiss · incremental-update · provenance · obsidian-export · mcp | Hyper-Extract is a Python CLI and library that turns unstructured documents into structured knowledge — graphs, hypergraphs, temporal and spatial graphs — using LLM extraction engines like GraphRAG, LightRAG and Hyper-RAG, with FAISS-backed semantic search, incremental provenance tracking and an optional MCP server. | View Project → |
| Codex-X | ★ 4K | AI Coding | Rust | codex · prompt-management · mcp · skills · provider-switching · session-management · config-editor · tauri · desktop-app · token-usage · toml · cc-switch · rust · cross-platform | Cross-platform desktop GUI (Tauri 2 + React + Rust) for managing OpenAI Codex CLI/desktop: prompt template injection, provider/API switching, session sync, Skills/MCP management, and TOML/auth.json editing with token usage stats. | View Project → |
| nano-graphrag | ★ 4K | RAG | Python | graphrag · rag · knowledge-graph · retrieval-augmented-generation · entity-extraction · community-detection · graph-query · llm · async · incremental-insert · neo4j · networkx · vector-database · hackable · small-codebase · ollama | nano-graphrag is a compact (~1100 LOC) MIT-licensed Python GraphRAG implementation: inserts text, builds an entity/relation graph, and answers global or local queries. Pluggable LLMs, embeddings, vector stores (FAISS, Milvus, Qdrant, hnswlib) and graph stores (networkx, Neo4j), fully async. | View Project → |
| rust-sdk | ★ 4K | MCP | Rust | MCP · Rust SDK · client · server · tools · resources · prompts · async | Official Rust SDK for implementing MCP clients and servers. Provides async protocol support, transports, typed tools and resources, and procedural macros for generating tool implementations. | View Project → |
| fast-graphrag | ★ 4K | RAG | Python | GraphRAG · knowledge graphs · retrieval · PageRank · incremental indexing · async · entity extraction · citations | Python GraphRAG library that builds and incrementally updates domain-specific knowledge graphs, then uses PageRank-based exploration to retrieve evidence for queries. Supports OpenAI-compatible models, Gemini, and Vertex AI. | View Project → |
| ghidra-mcp | ★ 4K | MCP | Java | mcp · mcp-server · ghidra · reverse-engineering · binary-analysis · static-analysis · decompilation · p-code · emulation · debugger · docker · headless · java · python · batch-operations · ai-agents | Ghidra MCP server exposing 253 tools that let AI agents drive Ghidra: decompile, rename, retype, document, emulate P-code, and live-debug binaries via GUI plugin or headless/Docker server. | View Project → |
| fast-agent | ★ 3.9K | AI Frameworks | Python | agents · mcp · mcp-client · mcp-server · a2a · acp · agent-skills · cli · tui · multi-agent · workflows · structured-outputs · multimodal · evaluation · ollama · python | Python framework and CLI for building, running and evaluating LLM agents and workflows, with first-class MCP (client/server, sampling, elicitations), Agent Skills, ACP and A2A support. Includes a TUI coding agent plus declarative agent/workflow definitions and broad model provider coverage. | View Project → |
| Model-Optimizer | ★ 3.9K | AI Tools | Python | model optimization · quantization · post-training quantization · quantization-aware training · pruning · distillation · speculative decoding · sparsity · neural architecture search · inference acceleration · model compression | Python library for optimizing PyTorch, Hugging Face, and ONNX models with quantization, pruning, distillation, sparsity, NAS, and speculative decoding. Exports optimized checkpoints for deployment with TensorRT-LLM, TensorRT, vLLM, and SGLang. | View Project → |
| LazyLLM | ★ 3.9K | AI Frameworks | Python | multi-agent · llm · low-code · rag · fine-tuning · model-deployment · workflow-orchestration · pipeline · agents · tool-calling · vector-database · multimodal · inference · cli | LazyLLM is a Python low-code framework for assembling multi-agent LLM apps from modular pipelines (pipeline, parallel, switch, loop). It bundles RAG, tool-calling agents, one-click deployment, and unified online/local model fine-tuning and inference via vLLM, LightLLM or LMDeploy. | View Project → |
| notfair-plugin | ★ 3.9K | Skills | TypeScript | seo · geo · aeo · marketing · agent-skills · claude-code-plugin · mcp · paid-ads · google-ads · meta-ads · analytics · content-marketing · wordpress · gohighlevel · multi-agent · claude-skills | MIT-licensed plugin of 48 SKILL.md marketing skills for Claude Code, Codex, Cursor, and Gemini CLI, covering SEO, GEO, paid ads, GA4/Search Console, WordPress, and CRM via one hosted NotFair MCP connection. | View Project → |
| cumora | ★ 3.9K | AI Agents | TypeScript | multi-agent · team-chat · ai-agents · byoa · agent-coordination · claude-code · codex · electron · cross-platform · kanban · agent-memory · realtime · postgres · redis · kubernetes · llm-cost-tracking | Cumora is cross-platform team chat where AI agents are first-class teammates alongside humans. Agents share DMs, group chats, Kanban and calendar; they run on Cumora Cloud or BYOA using Claude Code, Codex, or other local agent engines. | View Project → |
| stable-audio-tools | ★ 3.9K | AI Frameworks | Python | audio-generation · text-to-audio · diffusion-models · latent-diffusion · generative-audio · training · inference · autoencoder · fine-tuning · lora · audio-codec · music-generation · conditioning · gradio · deep-learning · checkpoints | Stability AI's toolkit for training and running generative audio models: latent diffusion, autoencoders, and LMs with text/audio conditioning. Includes training scripts, fine-tuning/LoRA support, and a Gradio UI for Stable Audio Open. | View Project → |
| kagent | ★ 3.9K | AI Agents | Go | kubernetes · cloud-native · agents · mcp · devops · cncf · declarative · controller · cli · web-ui · observability · opentelemetry · a2a · helm · llm-providers · tool-servers | Kubernetes-native framework for building, deploying and managing AI agents as custom resources. Agents combine a system prompt, MCP tool servers, and LLM ModelConfig (OpenAI, Anthropic, Vertex, Ollama), with a controller, UI, CLI, and an ADK-based engine. | View Project → |
| openscience | ★ 3.8K | AI Agents | TypeScript | scientific research · research agent · multi-agent · CLI · desktop app · browser workspace · scientific computing · data analysis · literature search · agent skills · MCP · local models | An AI research agent with CLI, desktop, and browser workbenches. It searches scientific sources, writes and runs analysis code and experiments, delegates bounded tasks to specialist workers, and records its steps and outputs for review. | View Project → |
| lean-ctx | ★ 3.8K | AI Coding | Rust | context-engineering · token-optimization · mcp-server · ai-coding · coding-agents · context-compression · prompt-cache · session-memory · knowledge-graph · rust · local-first · lsp · tree-sitter · cost-tracking · developer-tools · context-layer | Local-first context intelligence layer and MCP server for AI coding agents: it compresses file reads and shell output, routes and remembers context across sessions, and tracks token savings, cost and accepted outcomes via a live dashboard. | View Project → |
| DSH-better-sidebar | ★ 3.8K | AI Coding | TypeScript | DSH · sidebar · AI coding · side chat · subagents · file editor · Git · plugin extensions | A DSH plugin that adds a coding workspace with file browsing and editing, Git and AI-change views, subagent monitoring, and contextual side chats. It also exposes an API for other plugins to register sidebar tabs and file viewers. | View Project → |
| SimpleMem | ★ 3.8K | RAG | Python | long-term-memory · llm-agents · memory · rag · multimodal · semantic-search · retrieval · embedding · compression · knowledge-graph · mcp · cross-session-memory · audio · video · vision · self-evolving | SimpleMem is a lifelong memory stack for LLM agents: it stores dialogues and multimodal inputs as compressed atomic memories with embeddings, then retrieves them semantically. Ships a Python package, MCP server, cross-session memory, and a self-evolving retrieval tuner. MIT, Python. | View Project → |
| atmosphere | ★ 3.8K | AI Agents | Java | java · jvm · ai-agents · multi-agent · mcp · a2a · ag-ui · spring-ai · langchain4j · websocket · sse · grpc · token-streaming · human-in-the-loop · governance · durable-sessions | Atmosphere is a JVM runtime for stateful AI agents: one @Agent class streams tokens and tool calls over WebSocket/SSE/gRPC and runs on Spring AI, LangChain4j, Anthropic, or other adapters via a single AgentRuntime SPI. It packages multi-agent coordination, MCP/A2A/AG-UI endpoints, human approvals, memory, and governanc | View Project → |
| guppylm | ★ 3.8K | AI Frameworks | Python | llm · tiny-llm · transformer · from-scratch · training · tokenizer · bpe · onnx · wasm · education · pytorch · synthetic-data · inference · colab · quantization | GuppyLM is a ~9M parameter vanilla transformer trained from scratch to chat as a fish persona. It includes data generation, BPE tokenizer training, training loop, inference, ONNX/WASM browser demo, and Colab notebooks, making it a compact reference for building your own tiny LLM. | View Project → |
| zvec-grep | ★ 3.8K | MCP | Rust | local-first · hybrid-search · bm25 · vector-search · semantic-search · code-search · ripgrep · full-text-search · cli · embeddings · mcp-server · workspace-indexing · agent-context · code-navigation · source-citation · multi-format-search | zg is a local-first hybrid workspace search tool and MCP server that unifies ripgrep, BM25 and vector/semantic search over code and documents. It gives humans a CLI and agents a tool for ranked, source-linked results with fewer tokens and tool calls. | View Project → |
| open-higgsfield | ★ 3.8K | AI Tools | TypeScript | image-generation · video-generation · generative-ai · studio-ui · prompt-composer · gallery · batch-generation · self-hosted · nextjs · zustand · vercel-blob · indexeddb · model-catalog · media-uploads · open-source · byok | Open-source self-hosted studio for image and video generation: one prompt bar drives 38 models (8 image, 30 video) with per-model settings, media inputs, batch runs, and a gallery with history, reuse, and bulk actions. Uses your own platform key via server actions. | View Project → |
| pipeshub-ai | ★ 3.8K | RAG | Python | enterprise-search · rag · permission-aware · knowledge-graph · graphrag · connectors · mcp · self-hosted · citations · multi-agent · langchain · langgraph · notion · slack · ollama · deep-research | PipesHub is a self-hostable enterprise knowledge platform that connects 50+ business systems to AI. It offers permission-aware search with verified citations, GraphRAG retrieval, and exposes the same context to agents via MCP and Python/TS/Go SDKs. | View Project → |
| hyperresearch | ★ 3.7K | AI Agents | Python | deep research · research agent · Claude Code · OpenAI Codex · agent skills · web research · source provenance · citation verification · persistent knowledge base · multi-agent · adversarial review · research pipeline | A deep-research agent pipeline for Claude Code and OpenAI Codex that searches the web and scholarly sources, checks claims and citations, and saves sources in a persistent searchable vault. Includes agent skills, parallel research roles, and resumable runs. | View Project → |
| Financial-API | ★ 3.7K | MCP | TypeScript | a-share · financial-data · mcp · cli · python-sdk · agent-skill · rest-api · quantitative-finance · duckdb · stock-market · china-stock-market · market-data · npm-package · agent-integration | Official HiThink/Tonghuashun A-share financial data service: real-time and historical quotes, financials, indices, sectors, funds, futures and options via REST API, six hosted MCP servers, Node CLI, Python SDK and an installable hithink-finance Agent Skill. Requires an API key. | View Project → |
| mcp | ★ 3.7K | MCP | C# | mcp · model-context-protocol · microsoft · azure · mcp-servers · dotnet · csharp · ai-agents · cloud-infrastructure · fabric · developer-tools · enterprise · local-server · remote-server · tool-integration · monorepo | Microsoft's monorepo of official MCP server implementations, including Azure MCP, Microsoft Fabric MCP, Foundry, and engineering libraries/pipelines. Use it to give AI agents and IDE assistants standardized, one-click tool access to Microsoft cloud and developer services. | View Project → |
| Agentic-Design-Patterns | ★ 3.7K | AI Agents | Jupyter Notebook | AI agents · agentic design patterns · multi-agent systems · prompt chaining · tool use · planning · agent memory · MCP · RAG · guardrails · agent evaluation · Jupyter notebooks | An educational guide with a book PDF and runnable Jupyter notebooks covering 21 AI agent design patterns, from prompt chaining and tool use to multi-agent coordination, MCP, RAG, guardrails, and evaluation. | View Project → |
| SemIf | ★ 3.6K | AI Tools | Python | semantic-decisions · logit-readout · llm-inference · classification · benchmarking · agents · webgpu · gguf · quantization · evaluation · probability-calibration · reranking · pytorch · reproducibility · local-gpu · runtime-criteria | SemIf reads typed option probabilities directly from open LLM logits to answer runtime-defined decisions (routing, retries, evidence checks) without generating or parsing text. Includes Python CLI, frozen benchmarks, and a WebGPU browser demo on 4B-class models. | View Project → |
| goclaw | ★ 3.6K | AI Agents | Go | multi-agent · agent-orchestration · AI gateway · multi-tenant · MCP · memory · knowledge vault · Discord · Telegram · Slack · security · self-hosted | A Go-based, multi-tenant AI agent platform with agent teams, delegated tasks, persistent memory, a pgvector-backed knowledge vault, and integrations for messaging channels and MCP. Supports 20+ LLM providers and includes a desktop edition. | View Project → |
| linkedin-mcp-server | ★ 3.6K | MCP | Python | mcp · mcp-server · model-context-protocol · linkedin · browser-automation · scraping · claude · python · fastmcp · patchright · docker · job-search · messaging · recruiting · networking · stdio-http-transport | MCP server that gives Claude and other MCP clients LinkedIn access through your own logged-in browser session: profiles, companies, jobs, messages, connections, feed and post search. Runs locally with stdio or streamable HTTP transport, Docker and uvx installs. | View Project → |
| everything-claude-code | ★ 3.6K | AI Coding | JavaScript | claude-code · claude-code-plugin · subagents · slash-commands · hooks · rules · prompt-pack · developer-tooling · tdd · code-review · context-management · session-memory · evals · verification-loops · cross-platform · workflow-automation | Claude Code plugin bundling production-ready subagents, slash commands, skills, hooks, rules, and MCP configs for AI-assisted development. Covers TDD, code/security review, E2E testing, evals, session memory persistence, token optimization, and subagent orchestration. | View Project → |
| Observal | ★ 3.6K | AI Tools | Python | mcp · agents · skills · registry · self-hosted · observability · insights · cli · session-replay · governance · llm-ops · claude-code · cursor · codex · hooks · prompts | Self-hosted control plane and registry for internal AI components: bundle skills, MCP servers, hooks, prompts and sandboxes into versioned agents, install them into Claude Code/Cursor/Codex and other harnesses, and get insight reports plus session replay from real usage data. | View Project → |
| dbhub | ★ 3.6K | MCP | TypeScript | mcp · mcp-server · database · sql · postgres · mysql · sqlite · sql-server · oracle · mariadb · token-efficient · claude · cursor · ssh-tunnel · read-only · typescript | DBHub is a minimal, token-efficient MCP server exposing PostgreSQL, MySQL, MariaDB, SQL Server, Oracle and SQLite through two default tools (execute_sql, search_objects). It supports multi-connection TOML config, SSH tunneling, read-only guardrails, and works with Claude Code, Cursor and VS Code. | View Project → |
| knowhere | ★ 3.5K | RAG | Python | rag · agentic-rag · document-parsing · document-intelligence · chunking · vision-model · ocr · knowledge-base · retrieval · citations · pdf · pptx · mcp · vector-database · memory · long-context | Knowhere is an open-source document parsing and retrieval backend that converts PDFs, Office files, images, and text into hierarchy-native chunks with citations and cross-document links. It serves this memory to AI agents and RAG pipelines, including via MCP. | View Project → |
| antigravity-sdk-python | ★ 3.5K | AI Agents | Python | AI agents · Google Antigravity · Gemini · MCP · tool calling · streaming · multimodal · local models · agent policies · Python SDK | Python SDK for building stateful AI agents with Google Antigravity and Gemini. Includes streaming, custom tools, MCP server integration, policy hooks, multimodal inputs, and support for local Gemma models through LiteRT-LM. | View Project → |
| dembrandt | ★ 3.5K | MCP | TypeScript | design-tokens · mcp · mcp-server · cli · playwright · design-systems · design-to-code · wcag · tailwind-css · shadcn-ui · dtcg · ci · token-drift · typography · web-scraping · typescript | CLI and MCP server that renders a site with Playwright and extracts its design system (colors, typography, spacing, borders, shadows, motion, components) as terminal output, JSON, W3C DTCG, Tailwind, shadcn/ui or DESIGN.md. Also compares against a baseline in CI to fail on token drift. | View Project → |
| buildwithclaude | ★ 3.5K | Skills | Python | claude-code · plugin-marketplace · claude-skills · subagents · slash-commands · hooks · mcp-servers · registry · discovery · claude-desktop · agent-sdk · openclaw · cli · nextjs | Curated plugin marketplace and web discovery hub for Claude Code extensions: 117 agents, 175 commands, 28 hooks, 26 skills, plus indexing of 20k+ community plugins, MCP servers and marketplaces. Python/Node tooling validates manifests and powers the Next.js browse UI. | View Project → |
| radar | ★ 3.5K | MCP | Go | kubernetes · mcp · mcp-server · model-context-protocol · ai-agents · kubernetes-ui · kubernetes-troubleshooting · gitops · helm · argocd · fluxcd · observability · topology · kubectl-plugin · self-hosted · go | Single-binary Kubernetes UI for debugging clusters (topology, events, Helm, GitOps, traffic, audits) with a built-in MCP server so AI agents can inspect and operate the cluster. Go + web UI, no cluster-side install required. | View Project → |
| evalscope | ★ 3.5K | AI Tools | Python | LLM evaluation · benchmarking · performance testing · VLM evaluation · RAG evaluation · agent evaluation · multimodal · model comparison · evaluation reports · load testing | EvalScope is a Python framework for evaluating LLM, vision-language, embedding, reranking, and generative models. It includes benchmark integrations, RAG and agent evaluation, inference load testing, and a dashboard for comparing results. | View Project → |
| Vibe-Skills | ★ 3.5K | Skills | Python | agent-skills · skill-routing · workflow-orchestration · skill-discovery · context-engineering · prompt-engineering · multi-agent · task-decomposition · harness · claude-code · codex · vibe-coding · developer-tools · automation · verification-gates · token-efficiency | A general-purpose Skill for AI agents that discovers and routes installed local Skills, then decomposes complex tasks into dependency-ordered work units. Benchmarked on SkillsBench with +21.12pp reward and ~30% fewer tokens vs baseline. | View Project → |
| jianying-editor-skill | ★ 3.5K | Skills | Python | agent skill · video editing · Jianying · CapCut China · video automation · timeline generation · text-to-speech · subtitles · screen recording | An AI agent skill that automates Jianying desktop editing by building project timelines, arranging media, generating voiceovers and subtitles, and applying effects. Supports Windows and macOS draft creation; automated export is limited to Windows. | View Project → |
| mcp-server-browserbase | ★ 3.4K | MCP | TypeScript | mcp · browser-automation · browserbase · stagehand · playwright · web-scraping · computer-use · typescript · stdio · http-transport · archived · cloud-browser · session-management · data-extraction | Browserbase MCP server exposing 6 tools (start, end, navigate, act, observe, extract) so LLMs can control a cloud browser via Browserbase and Stagehand. Supports STDIO and HTTP transports, custom models, proxies and persistent contexts. Repo is archived and no longer maintained. | View Project → |
| wewrite | ★ 3.4K | Skills | Python | wechat · 公众号 · agent-skill · content-creation · writing · markdown · publishing · image-generation · seo · multi-platform · style-learning · topic-selection · claude-code · editorial-review · html-themes · cli | 给 AI Agent（Claude Code/Codex/OpenClaw/Hermes）用的公众号内容全流程 Skill：从热点抓取、选题评分、素材核对、写作审稿，到 18+ 主题排版、配图与推送草稿箱。prompt 负责判断，Python CLI 负责确定性操作，状态存于 ~/.wewrite/。 | View Project → |
| metorial | ★ 3.4K | MCP | TypeScript | MCP · AI agents · integrations · OAuth · access control · RBAC · audit logs · tool calling · security | Metorial provides an access and integration layer for AI agents, with 1,200+ integrations and MCP support. It centralizes authentication, permissions, and audit logging, and connects to agent frameworks through SDKs. | View Project → |
| effective-html | ★ 3.3K | Skills | HTML | agent-skills · HTML artifacts · wireframes · prototypes · diagrams · plans · responsive design · accessibility | A collection of reusable agent skills for creating self-contained HTML artifacts, including wireframes, interactive prototypes, plans, and diagrams. Supports Claude Code and Codex, with guidance for responsive, accessible output. | View Project → |
| SocratiCode | ★ 3.3K | MCP | TypeScript | mcp · mcp-server · semantic-search · code-graph · codebase-intelligence · embeddings · vector-database · qdrant · ollama · hybrid-search · impact-analysis · ast · local-first · private · docker · claude-code | SocratiCode is a zero-config, local-first MCP server (plus Claude Code/Cursor/VS Code/Gemini plugins) that indexes large codebases via Docker, Qdrant and Ollama for hybrid semantic+BM25 search, polyglot dependency graphs, and symbol-level impact/call-flow analysis. No API keys required; data stays on your machine. | View Project → |
| Sidekick | ★ 3.3K | AI Agents | Swift | local LLM · AI agent · RAG · deep research · macOS · offline AI · web search · function calling · file search · image generation | A native macOS AI assistant that runs local models with llama.cpp and answers questions grounded in files, folders, and websites. Includes RAG, web search, function-calling agents, deep research, memory, and support for OpenAI-compatible APIs. | View Project → |
| sie | ★ 3.3K | AI Frameworks | Python | inference server · model serving · agent infrastructure · embeddings · reranking · semantic search · RAG · OpenAI-compatible API · self-hosted · Kubernetes · multimodal | Self-hosted inference server and production cluster for agent workloads. Serves embedding, retrieval, reranking, generation, OCR, extraction, and multimodal models through an OpenAI-compatible API, with on-demand model loading and integrations for popular AI frameworks and vector stores. | View Project → |
| ultracite | ★ 3.3K | AI Coding | TypeScript | AI coding · agentic development · linter · formatter · zero-config · TypeScript · monorepo · Claude Code · Codex | A zero-configuration linting and formatting setup for JavaScript and TypeScript projects, with presets for Biome, ESLint, and Oxlint. It can generate shared rules and context files for AI agents and hand unresolved fixes to Claude Code or Codex. | View Project → |
| Clypra | ★ 3.3K | AI Tools | TypeScript | video-editor · tauri-v2 · react-19 · rust · typescript · wgpu · ffmpeg · gpu-acceleration · timeline · desktop-app · cross-platform · capacitor · media-tooling · open-source · mit-license | Clypra is a free MIT-licensed desktop video editor built with Rust, Tauri v2, and React 19, using FFmpeg plus hardware GPU decoding (VideoToolbox/D3D11VA/VAAPI) for sub-10ms frame decode and a frame-accurate multi-track timeline. It is a media/creative tool, not an AI, agent, or LLM project. | View Project → |
| neo | ★ 3.3K | AI Frameworks | JavaScript | javascript · frontend-framework · multi-threaded · web-workers · zero-build · es-modules · json-first-ui · vdom · agent-memory · graph-rag · mcp-server · multi-agent · self-healing · knowledge-graph · nodejs | Neo.mjs is a multi-threaded JavaScript application engine (worker-based runtime, JSON-first UI, zero-build ES modules), framed as the runtime 'body' hosting an AI agent swarm; the Agent OS, MCP servers and GraphRAG live mostly in the sibling neo-agent-brain repo. | View Project → |
| fastmcp | ★ 3.3K | MCP | TypeScript | mcp · mcp-server · typescript · sse · http-streaming · oauth · openapi · sessions · edge-runtime · stdio · authentication · hono · zod · cli · server-framework · stateless-mode | FastMCP is a TypeScript framework built on the official MCP SDK that removes boilerplate for defining tools, resources, and prompts, with auth, OpenAPI conversion, HTTP streaming/SSE, and edge/serverless support. Note: it targets legacy handshake MCP revisions, not the newer stateless spec. | View Project → |
| google_workspace_mcp | ★ 3.2K | MCP | Python | mcp · mcp-server · google-workspace · gmail · google-calendar · google-drive · google-docs · google-sheets · google-slides · oauth21 · cli · claude · chatgpt · multi-user · streamable-http · python | Production-grade MCP server exposing 120+ tools across all twelve Google Workspace services (Gmail, Drive, Calendar, Docs, Sheets, Slides, Forms, Tasks, Contacts, Chat, Search, Apps Script). Supports OAuth 2.1 multi-user auth, streamable HTTP/stdio, stateless containers, and ships a CLI for Claude Code and similar agen | View Project → |
| Switchyard | ★ 3.2K | AI Tools | Rust | llm-routing · model-routing · llm-gateway · openai-compatible · anthropic-compatible · cost-optimization · proxy-server · litellm · provider-abstraction · routing-algorithms · rust · python · benchmarking · prometheus-metrics · claude-code · codex-cli | NVIDIA Switchyard is a Rust+Python LLM routing layer that picks the cheapest capable model per call while staying OpenAI/Anthropic API compatible. It works as a NeMo Relay plugin, a LiteLLM routing plugin, an embeddable library (switchyard-libsy), or a standalone proxy that Claude Code and Codex CLI can point at. | View Project → |
| harbor | ★ 3.2K | AI Tools | Python | llm-stack · docker-compose · local-llm · self-hosted · cli · orchestration · inference · open-webui · ollama · llamacpp · vllm · mcp · llm-ops · homelab · agents | Harbor is a CLI and companion app that boots a fully pre-wired local LLM stack via Docker Compose. One command starts backends (Ollama, llama.cpp, vLLM, MLX), frontends (Open WebUI, LibreChat), and services like SearXNG, Speaches, and ComfyUI, with cross-service config handled for you. | View Project → |
| bazi-skill | ★ 3.2K | Skills | Python | claude-code-skill · agentskills · bazi · four-pillars · chinese-astrology · metaphysics · divination · python · skill-pack · calendar-conversion · wuxing · ten-gods · dayun · shensha · stdlib-only | A Claude Code Agent Skill for Four Pillars (Bazi) Chinese astrology. It interactively collects birth info, computes year/month/day/hour pillars plus luck cycles with a stdlib-only Python script, and analyzes them against nine classical texts. | View Project → |
| WorkBuddyGuide | ★ 3.2K | Skills | TypeScript | workbuddy · agent-skills · mcp · multi-agent · automation · workflow · guide · documentation · tutorial · llm · prompt-templates · connectors · vitepress · chinese · case-studies · knowledge-base | Community-maintained open-source bluebook for WorkBuddy: tutorial chapters, real case studies, and prompt templates covering Skills, MCP connectors, automation, and multi-agent workflows. It is a VitePress docs site (Cloudflare Pages), not runnable agent code. | View Project → |
| cortex | ★ 3.2K | MCP | TypeScript | mcp-server · openapi · asyncapi · graphql · grpc · openrpc · sdk-generator · code-generation · api-documentation · api-reference · markdown · cli · typescript · multi-language · static-docs · agents | Cortex turns OpenAPI, AsyncAPI, GraphQL, gRPC and OpenRPC specs plus Markdown into typed SDKs, static interactive docs, and an MCP server that exposes specs, SDK guides and project docs as agent tools. One config drives all outputs. | View Project → |
| fli | ★ 3.2K | MCP | Python | mcp · mcp-server · google-flights · flights · travel · flight-search · cli · python-library · fastmcp · stdio · http-streamable · api-wrapper · rate-limiting · pydantic | Fli is a Python library, CLI, and MCP server that queries Google Flights' API directly (no scraping). It exposes search_flights and search_dates tools over STDIO or HTTP so LLM agents can look up fares, dates, and filters. | View Project → |
| chinese-novelist-skill | ★ 3.2K | Skills | Python | claude-code-skill · creative-writing · chinese-novel · webnovel · long-form-generation · prompt-engineering · agent-skill · memory · outline-generation · auto-validation · multi-agent-writing · resume-writing · llm-writing · writing-assistant · chapter-wordcount-check | A Claude Code agent skill that writes full Chinese novels (10-50 chapters): three-layer Q&A, preference memory, resume support, serial/sub-agent/team writing modes, and automatic wordcount and coherence validation with rewrite. | View Project → |
| OGAM | ★ 3.1K | AI Tools | TypeScript | offline-ai · on-device · gguf · llama.cpp · whisper · stable-diffusion · mcp · rag · react-native · privacy-first · tool-calling · vision-language-model · android · ios · macos · local-llm | Off Grid AI is a React Native app that runs GGUF LLMs, Whisper STT, Stable Diffusion image gen, vision models, and on-device RAG fully offline on Android, iOS, and Apple Silicon Macs, with MCP tool calling and local-network LLM server support. | View Project → |
| davinci-resolve-mcp | ★ 3.1K | MCP | Python | mcp · mcp-server · davinci-resolve · video-editing · post-production · blackmagic-design · color-grading · fairlight · fusion · model-context-protocol · python · node · render-automation · timeline-editing · control-panel · media-management | MCP server exposing DaVinci Resolve Studio's scripting API to AI assistants: 37 compound tools (389 granular) for editing, media pool, render, grading, Fusion and Fairlight, plus an optional offline Node server that edits .drp/.drt/.drx files with no Resolve running. | View Project → |
| Bonsai-demo | ★ 3.1K | AI Tools | Shell | local LLM · small models · quantized models · ternary weights · 1-bit weights · vision-language · tool calling · MCP · offline inference · OpenAI-compatible API | Scripts and setup files for running Bonsai, Ternary-Bonsai, and Bonsai 2 models locally with llama.cpp or MLX. Includes Open WebUI integration, vision support, tool calling, and MCP connections. | View Project → |
| seekdb | ★ 3.1K | RAG | C++ | vector-database · hybrid-search · rag · agent-memory · full-text-search · hnsw · mysql-compatible · embedded-database · copy-on-write · sql · semantic-search · acid · vector-index · state-store · sandbox · python-sdk | MySQL-compatible embedded/server database unifying vector, full-text and relational data in one engine. Two-level HNSW hybrid search, async index pipeline for streaming writes, and FORK/MERGE copy-on-write sandboxes make it a state store for agent memory and RAG. | View Project → |
| agent-scan | ★ 3.1K | AI Tools | Python | security · supply-chain-security · mcp · prompt-injection · agent-skills · scanner · cli · llm-security · ai-security · static-analysis · secrets-detection · tool-poisoning · vulnerability-scanning · python | Snyk Agent Scan is a Python CLI that discovers installed AI agent components (agents, MCP servers, and skills) on a machine and scans them for prompt injection, tool poisoning, malware payloads, and secret leakage. It auto-detects configs from Claude, Cursor, Copilot, Windsurf, Gemini CLI, and others. | View Project → |
| handraw-style | ★ 3.1K | Skills | HTML | prompt-library · image-generation · hand-drawn · agent-skill · codex · bilingual · prompt-engineering · gallery · layout-templates · color-palettes · comic-storyboard · infographic · social-card · poster-design · illustration · static-assets | A Codex/agent Skill packaging 278 numbered hand-drawn illustration styles, 119 layout templates and 30 monochrome theme colors into bilingual CN/EN image prompts. Includes a model-capability matrix for gpt-image-2 plus reference-image fallback for Midjourney, Flux and others. | View Project → |
| jarvis-registry | ★ 3K | MCP | Python | mcp · mcp-gateway · agent-gateway · a2a · agent-orchestration · registry · oauth2 · oidc · rbac · acl · observability · opentelemetry · prometheus · semantic-search · federation · enterprise-ai | Enterprise MCP and A2A agent gateway that fronts AI copilots and autonomous agents with OAuth2/OIDC identity, tool-level ACLs, skill-based discovery, and OpenTelemetry/Prometheus observability. Runs via Docker Compose with a Vue/TS admin UI. | View Project → |
| skills | ★ 3K | Skills | TypeScript | agent-skills · skills · mcp · ai-coding-agents · azure · microsoft-foundry · agents-md · custom-agents · github-copilot · plugins · evals · context-engineering · sdk · typescript · prompt-packs · copilot-cli | Microsoft's repo of 175 agent skills, custom agents, AGENTS.md templates and MCP configs that ground AI coding agents (Copilot CLI, VS Code, Claude, OpenCode) in Azure SDK and Microsoft Foundry patterns. Install selectively via `npx skills add microsoft/skills`. | View Project → |
| deepteam | ★ 3K | AI Tools | Python | red-teaming · llm-security · ai-safety · guardrails · jailbreak · prompt-injection · llm-evaluation · adversarial-attacks · vulnerability-scanning · ai-agents · penetration-testing · bias-detection · multi-turn · python · llm-testing · responsible-ai | DeepTeam is an open-source Python framework for red teaming LLM apps and AI agents. It ships 50+ vulnerability checks and 20+ adversarial attack methods, runs locally on top of DeepEval, and adds optional guardrails for production use. | View Project → |
| memobase | ★ 2.9K | RAG | Python | long-term-memory · user-profile · llm-memory · chatbot · rag · retrieval · embeddings · personalization · temporal-memory · fastapi · postgres · redis · python-sdk · mcp · batch-processing · memory-layer | Memobase is a user-profile-based long-term memory service for LLM apps: insert chat blobs, it batch-extracts structured profiles plus a time-aware event timeline, then returns prompt-ready context in <100ms. Self-hosted on FastAPI/Postgres/Redis with Python, Node, Go SDKs and an MCP server. | View Project → |
| skills | ★ 2.9K | Skills | Shell | agent skills · Cloudflare · Workers · Agents SDK · Durable Objects · Wrangler · MCP · developer guidance | A collection of agent skills that gives coding assistants Cloudflare-specific guidance for Workers, the Agents SDK, and related platform products. The plugin also bundles Cloudflare's remote MCP server for API and documentation access. | View Project → |
| genaiscript | ★ 2.9K | AI Frameworks | TypeScript | prompt-as-code · llm-orchestration · agents · mcp · rag · vector-search · tool-calling · evals · vscode-extension · cli · javascript · typescript · code-interpreter · content-safety · promptfoo · deprecated | Microsoft GenAIScript is a JavaScript/TypeScript framework for writing LLM prompts as code, orchestrating models, tools, MCP servers and agents. It includes built-in RAG vector search, structured output schemas, evals, and a VS Code extension plus CLI. Note: the repository is marked DEPRECATED. | View Project → |
| oh-my-hermes | ★ 2.9K | Skills | Python | hermes-agent · agent-skills · skill-pack · cli · workflow-automation · model-routing · long-term-memory · coding-agent · parallel-execution · prompt-calibration · evidence-gates · plugin · multi-surface-install · cost-optimization · developer-tooling · python | Plugin for Nous Research's Hermes Agent that adds a governed operating layer: complexity-based model routing, per-model prompt calibration, parallel worktree execution, and long-term memory. Ships an omh CLI plus skill packs that install into Claude Code, Codex, Cursor, and OpenCode surfaces. | View Project → |
| narrator-ai-cli-skill | ★ 2.9K | Skills | — | agent-skill · video-narration · video-generation · movie-commentary · content-creation · text-to-speech · voice-cloning · narrator-ai-cli | An agent skill that teaches compatible AI agents to operate narrator-ai-cli for end-to-end movie narration video creation, including script generation, voice selection, and video composition. Requires the CLI and an API key. | View Project → |
| treg | ★ 2.9K | AI Tools | Python | agent tools · tool registry · MCP · API proxy · credential injection · CLI · skills · team sharing · tool catalog · self-hosted | Treg is a self-hostable registry and proxy for agent tools, with a searchable catalog of third-party endpoints, team APIs, CLIs, and skills. It injects credentials server-side and exposes tools through a CLI and MCP endpoints. | View Project → |
| web-quality-skills | ★ 2.8K | Skills | Shell | agent-skills · claude-skills · claude-code-plugin · codex · gemini-cli · web-performance · lighthouse · core-web-vitals · accessibility · wcag · seo · best-practices · chrome-devtools-mcp · performance-traces · crux · measurement-first | Agent Skills pack giving coding agents a measurement-first workflow for auditing and fixing web quality: Lighthouse and DevTools traces, Core Web Vitals, WCAG 2.2 accessibility, SEO and best practices. Installs into Claude Code, Codex, Gemini CLI or Cursor via npx skills add. | View Project → |
| qwen-audio-agent | ★ 2.8K | AI Agents | JavaScript | voice agent · real-time audio · speech-to-text · text-to-speech · agent orchestration · ACP · multimodal · desktop app · WebUI · TUI · memory | A JavaScript runtime for real-time voice conversations with AI agents, offering WebUI, terminal, and desktop clients. It connects interchangeable voice services to backend agents such as Claude Code, Codex, and OpenCode, with parallel background tasks and session memory. | View Project → |
| tons-of-skills-marketplace | ★ 2.8K | Skills | Python | agent-skills · claude-code · plugin-marketplace · skills · mcp · marketplace · cli · package-manager · llm · ai-agents · automation · devops · developer-tools · plugin-system · prompt-packs · monorepo | Model-agnostic agent-skills marketplace: a harness-free canonical skill layer with verified Claude Code adapters, MCP presets, and the ccpi CLI for installing plugins and skills from a 434-plugin, 2,900-skill catalog. | View Project → |
| substrate | ★ 2.8K | AI Agents | Go | agent-runtime · sandboxing · kubernetes · microvm · gvisor · multiplexing · suspend-resume · zero-trust · kernel-isolation · orchestration · stateful-agents · go · serverless · infrastructure | Agent Substrate is a Go/Kubernetes runtime for high-density agent execution: it multiplexes stateful agent actors onto a small worker pool with sub-500ms suspend/resume, zero-trust kernel and network isolation, and support for microVM/gVisor sandboxes plus ADK, LangChain, MCP and coding agents. | View Project → |
Vibe Coding Discover