| agentic-inbox | ★ 8K | AI Agents | TypeScript | email · cloudflare-workers · ai-agent · mcp · durable-objects · self-hosted · workers-ai · email-routing · auto-draft · r2 · sqlite · hono · react-router · streaming · tool-calling · typescript | Self-hosted email client running on Cloudflare Workers with an embedded AI email agent (Cloudflare Agents SDK + Workers AI) that reads, searches and drafts replies, plus an MCP endpoint so external AI tools can act on mailboxes. | View Project → |
| osaurus | ★ 8K | AI Agents | Swift | macos · swift · apple-silicon · local-llm · offline · multi-agent · mcp-server · persistent-memory · sandbox · computer-use · browser-automation · privacy · image-generation · orchestrator · subagents · cryptographic-identity | Osaurus is a native Swift/SwiftUI macOS harness for AI agents: local or cloud models, persistent memory, sandboxed code execution, browser and computer use, image generation, MCP support, and cryptographic agent identity. Fully offline-capable and MIT licensed. | View Project → |
| munder-difflin | ★ 8K | AI Agents | TypeScript | multi-agent · agent-orchestration · local-first · claude-code · codex · gemini-cli · opencode · electron · desktop-app · agent-memory · hive-mind · autonomous-agents · mailbox-routing · terminal-agents · human-in-the-loop · typescript | Munder Difflin is a local-first Electron desktop harness that wraps terminal coding CLIs (Claude Code, Codex, Gemini CLI, Qwen, OpenCode, and more) into a coordinated office of agents with shared markdown memory, mailboxes, and a supervisor agent that routes work and escalates approvals. | View Project → |
| mini-swe-agent | ★ 7.9K | AI Coding | Python | coding-agent · swe-bench · cli · minimal · bash · agent · llm · sandboxing · python · open-source · benchmark · automation | mini-swe-agent is a ~100-line Python software engineering agent that solves GitHub issues and runs as a CLI coding assistant. It uses only bash, has no tool-calling scaffolding, supports any LiteLLM model, and scores >74% on SWE-bench verified. | View Project → |
| modly | ★ 7.9K | AI Tools | TypeScript | 3D generation · image-to-3D · local AI · GPU · desktop app · 3D meshes · workflows · self-hosted | Modly is a cross-platform desktop app for generating 3D meshes from images or prompts using AI models that run locally on your GPU. It supports configurable workflows, model extensions, mesh optimization, and a CLI for automation. | View Project → |
| rikkahub | ★ 7.8K | AI Tools | Kotlin | android · kotlin · llm-chat-client · multi-provider · mcp · chatbot-ui · multimodal · agent · jetpack-compose · material-you · web-ui · prompt-variables · message-branching · memory · search-integration · self-hosted | RikkaHub is a native Android LLM chat client with Material You UI that lets you switch between any OpenAI/Anthropic/Gemini-compatible provider. It adds MCP support, custom agents, memory, multimodal input, search grounding and a web UI. | View Project → |
| notebooklm-skill | ★ 7.8K | Skills | Python | claude-code · claude-skills · notebooklm · browser-automation · patchright · citation-backed · source-grounded · knowledge-base · document-qa · python · google-gemini · hallucination-reduction · library-management · persistent-auth · archived · local-only | Claude Code skill (Python + Patchright browser automation) that lets Claude query your Google NotebookLM notebooks for source-grounded, citation-backed Gemini answers. Local Claude Code only; repo archived Sept 2026 and no longer maintained. | View Project → |
| kev | ★ 7.8K | AI Frameworks | Python | decision-model · calibration · lora · qwen3 · inference-engine · serving · evals · benchmarks · probabilities · block-causal-mask · typesafe · fastapi · pytorch · playground · chess · text-classification | Kev trains LoRA adapters plus a pointer readout head on Qwen3 (0.6B/4B/8B) to answer typed yes/no, choice, and score questions over a document in one causal prefill pass, returning calibrated probabilities rather than text. Ships a FastAPI /v1/systemone server, frozen eval suites, and a Next.js playground. | View Project → |
| kimi-code | ★ 7.7K | AI Coding | TypeScript | cli · coding-agent · terminal · tui · mcp · subagents · acp · plugin-marketplace · hooks · video-input · typescript · monorepo | Kimi Code CLI is a terminal AI coding agent from Moonshot AI that reads and edits code, runs shell commands, searches files and fetches web pages. It ships as a single binary with a fast TUI, MCP config, plugins, subagents and ACP support for Zed/JetBrains. | View Project → |
| firecrawl-mcp-server | ★ 7.5K | MCP | JavaScript | mcp-server · web-scraping · web-search · crawler · content-extraction · javascript-rendering · batch-processing · deep-research · browser-interaction · page-monitoring · self-hosted · sse · streamable-http · oauth · typescript · llm-tools | Official Firecrawl MCP server that gives MCP-compatible clients (Cursor, Claude, VS Code, Windsurf) web search, scraping, crawling, page interaction, monitoring, and deep research over stdio, SSE, or streamable HTTP, with cloud, keyless, and self-hosted modes. | View Project → |
| openllmetry | ★ 7.4K | AI Tools | Python | observability · opentelemetry · llmops · tracing · monitoring · instrumentation · llm · metrics · telemetry · vector-db · python · eval · genai · apm | OpenLLMetry provides OpenTelemetry-based instrumentations for LLM providers, vector DBs and agent frameworks, plus a Python SDK that emits standard OTel traces and metrics. Use it to trace, monitor and export LLM app telemetry into Datadog, Grafana, Honeycomb and other backends. | View Project → |
| gentle-ai | ★ 7.3K | AI Coding | Go | ai-coding · coding-agents · claude-code · cursor · codex · opencode · mcp · agent-skills · persistent-memory · spec-driven-development · code-review · developer-tools · golang · personas · tdd · no-lock-in | Go CLI that configures the AI coding agents you already use (Claude Code, Cursor, Codex, OpenCode, plus 12 more) with persistent memory, Spec-Driven Development workflows, curated skills, MCP servers, personas and optional bounded code review, with no agent lock-in. | View Project → |
| Windows-MCP | ★ 7.3K | MCP | Python | Windows · computer use · desktop automation · UI automation · MCP server · keyboard and mouse · browser automation · file navigation | An MCP server that exposes Windows desktop controls and UI state to AI clients. Agents can launch apps, interact with windows, send input, capture screenshots, and automate browser pages. | View Project → |
| guizang-social-card-skill | ★ 7.3K | Skills | HTML | agent-skill · claude-code · codex · claude-skill · social-cards · xiaohongshu · rednote · wechat · html-to-png · playwright · editorial-design · swiss-design · live-photo · image-generation · layout-templates · carousel | A Claude Code / Codex agent skill that turns articles, copy, screenshots or videos into Xiaohongshu carousels, WeChat 21:9+1:1 cover pairs and Live Photo cards. Ships 28 layouts, 10 theme presets, editorial/Swiss style systems, single-file HTML rendered to PNG via Playwright. | View Project → |
| Yuxi | ★ 7.2K | AI Agents | Python | multi-agent · agentic-rag · knowledge-graph · multi-tenant · self-hosted · mcp · agent-skills · sandbox · knowledge-base · langgraph · document-parsing · rag-evaluation · subagents · rbac · kbqa · docker | Yuxi is a self-hosted, multi-tenant knowledge agent platform that combines Agentic RAG, Neo4j knowledge graphs, LangGraph multi-agent orchestration, MCP/Skills, sandboxed file tools and RBAC in one Docker-deployed workspace. Aimed at teams that need to own their data, models and permissions. | View Project → |
| swarms | ★ 7.2K | AI Agents | Python | multi-agent · agent-orchestration · swarms · mcp · workflows · sequential-workflow · concurrent-workflow · graph-workflow · group-chat · mixture-of-agents · agent-memory · tool-calling · prompt-engineering · production-agents · autonomous-agents · python | Python framework for orchestrating multi-agent systems: prebuilt sequential, concurrent, graph, and mixture-of-agents architectures, plus MCP client/server support so agents can call external tools or be exposed as MCP tools. Apache-2.0, pip-installable, runs on any LiteLLM-supported model. | View Project → |
| mlx-lm | ★ 7.2K | AI Frameworks | Python | llm-inference · apple-silicon · mlx · quantization · fine-tuning · lora · gguf · text-generation · prompt-caching · local-llm · openai-compatible · cli · chat-repl · model-conversion · distributed-inference · benchmarking | MLX LM is a Python package for running and fine-tuning LLMs on Apple silicon via MLX. It offers CLI and Python APIs for generation, chat, LoRA/full fine-tuning, quantization, GGUF conversion, prompt caching, and an OpenAI-compatible server. | View Project → |
| skills | ★ 7.2K | Skills | Markdown | agent-skills · ui-design · typography · color-system · accessibility · layout · design-engineering · product-writing · claude-code-plugin · design-review · component-variants · frontend · prompt-pack · skill-pack | Agent skill pack (also a Claude Code plugin, installable via npx skills) that improves frontend interfaces: UI polish, typography, color systems, accessibility, layout and product writing, plus review, explain, break and variant workflows. | View Project → |
| DeepSpec | ★ 7.2K | AI Frameworks | Python | speculative-decoding · llm-inference · inference-optimization · draft-model · model-training · eagle3 · dflash · dspark · pytorch · triton · distributed-training · llm-evaluation · throughput · latency · transformers · benchmarking | DeepSpec is a full-stack Python codebase for training and evaluating speculative-decoding draft models (DSpark, DFlash, Eagle3) against targets like Qwen3 and Gemma. It covers data prep, 8-GPU training, and benchmark evaluation, plus released checkpoints. | View Project → |
| stable-diffusion.cpp | ★ 7.2K | AI Tools | C++ | image-generation · video-generation · image-editing · diffusion-inference · local-inference · C++ · GGUF · quantization · CUDA · Vulkan · Metal · CLI | A C/C++ inference engine for local image and video diffusion models, with CLI generation, image editing, model conversion, and CPU, CUDA, Vulkan, Metal, OpenCL, and SYCL backends. | View Project → |
| awesome-agentic-ai-zh | ★ 7.1K | AI Agents | Python | agentic-ai · learning-roadmap · awesome-list · multi-agent-systems · mcp · rag · prompt-engineering · claude-code · agent-skills · cli-agents · education · tutorial · trilingual · chinese · llm · python | Trilingual (繁中/EN/简中) learning roadmap for agentic AI: 9 stages from LLM basics to multi-agent production, with 240+ curated resources, runnable Python starter examples and role-based tracks (developer, researcher, teacher, knowledge worker). | View Project → |
| OpenAlice | ★ 7.1K | AI Agents | TypeScript | trading · finance · multi-agent · agent-orchestration · market-data · broker-integration · ibkr · quantitative-research · prediction-markets · scheduled-tasks · local-first · electron · cli · desktop-app · workspaces · paper-trading | OpenAlice is a local AI trading agent orchestrator that connects agents like Claude Code and Codex to market data, research workspaces, scheduled tasks, and broker accounts. It covers research, thesis tracking, paper trading, and approved order execution across equities, crypto, forex, and macro. | View Project → |
| oh-story-claudecode | ★ 7.1K | Skills | Python | agent-skills · claude-code-skills · novel-writing · creative-writing · chinese-web-novel · writing-assistant · ai-writing · de-ai-flavor · story-outline · fiction-writing · claude-code · codex · opencode · continuity-tracking · cover-generation · prompt-pack | A 13-skill pack that turns coding agents (Claude Code, Codex, OpenCode, Antigravity and others) into a Chinese web-novel writing pipeline: chart scanning, novel deconstruction, outlining, drafting, AI-flavor removal and cover generation, with file-based continuity tracking for hundreds of chapters. | View Project → |
| vespa | ★ 7.1K | RAG | Java | AI search · vector search · vector database · RAG · hybrid search · tensor · machine learning · recommendation · personalization · distributed serving | Vespa is a distributed platform for indexing and searching text, vectors, tensors, and structured data, with machine-learning inference at serving time. Use it to build large-scale search, recommendation, personalization, and RAG systems. | View Project → |
| CyberStrikeAI | ★ 7.1K | AI Agents | Go | ai-agents · multi-agent · cybersecurity · penetration-testing · mcp · rag · security-tools · human-in-the-loop · attack-chain · red-team · agent-skills · ctf · vulnerability-management · c2 · orchestration · go | CyberStrikeAI is a Go-based AI-native security operations platform that orchestrates Eino-powered agents, 100+ MCP security tools, RAG knowledge, and human-in-the-loop approvals for authorized pentesting and red-team work, with attack-chain modeling, audit logs, and WebShell/C2 integrations. | View Project → |
| opensquilla | ★ 7.1K | AI Agents | Python | ai-agent · token-efficiency · model-router · mcp · persistent-memory · multi-channel · local-embeddings · sandbox · cli · web-ui · skills · scheduling · self-hosted · tool-use · llm-router · cost-optimization | OpenSquilla is a Python microkernel AI agent runtime: a local model router sends each turn to the cheapest capable LLM, with persistent memory, sandboxed tools, MCP-native integrations and one shared loop across CLI, Web UI and chat channels. Ships desktop/uv installs, skills, scheduling and 20+ provider support. | View Project → |
| ax | ★ 7K | AI Agents | Go | agent-orchestration · sandboxing · kubernetes · declarative · multi-agent · task-runner · mcp · workspaces · git · network-egress · suspend-resume · cli · grpc · cluster · go · checkpointing | Google's AX is a declarative, Kubernetes-style orchestrator for autonomous agent workloads. Apply Task/Workspace/Gateway/Model YAML to sandbox agents, wire Git and MCP servers, fence egress traffic, and suspend/resume state at cluster scale. | View Project → |
| repowise | ★ 7K | MCP | Python | mcp · code-intelligence · code-health · static-analysis · dead-code · git-analytics · dependency-graph · documentation-generator · refactoring · technical-debt · developer-tools · codebase-index · ast · ai-agents · self-hosted · python | Self-hosted MCP server that indexes a repo's dependency graph, git history, docs and architectural decisions, then serves task-shaped context to coding agents. Adds deterministic code-health scoring, dead-code and change-impact analysis, plus CLAUDE.md/AGENTS.md generation. | View Project → |
| mobile-mcp | ★ 7K | MCP | TypeScript | mobile automation · iOS · Android · device control · app testing · UI interaction · accessibility tree · screenshots · data extraction · emulators · simulators · real devices | An MCP server that lets compatible AI clients control iOS and Android apps on simulators, emulators, and real devices. Provides accessibility-based UI inspection, taps and gestures, app management, screenshots, device logs, and batch commands. | View Project → |
| gpt-load | ★ 7K | AI Tools | Go | AI gateway · LLM gateway · self-hosted · API routing · load balancing · failover · credential management · usage tracking · OpenAI-compatible · Anthropic · Gemini | A self-hosted gateway that unifies AI provider APIs and subscription accounts behind one endpoint. It provides credential scheduling, retries and failover, request logs, usage tracking, and a web management UI. | View Project → |
| ODS | ★ 6.8K | AI Tools | Python | local AI · self-hosted · AI server · local inference · RAG · AI agents · workflow automation · voice AI · image generation · Docker · privacy · homelab | ODS installs and configures a self-hosted AI stack on Linux, Windows, or macOS. It combines local inference, a chat UI, RAG, voice, agents, workflow automation, and image generation behind an installer and management dashboard. | View Project → |
| zotero-pdf2zh | ★ 6.8K | AI Tools | Python | pdf-translation · zotero-plugin · document-translation · llm-translation · bilingual-pdf · layout-preservation · formula-preservation · ocr · batch-translation · terminology-extraction · self-hosted · academic-papers · typescript-plugin · python-server · docker-deploy · translation-engine | Zotero plugin that translates academic PDFs into Chinese via LLM services (pdf2zh/pdf2zh_next) while preserving formulas and layout. Adds bilingual/crop reading, batch translation, OCR, terminology extraction, and a self-hosted Python server with web progress UI. | View Project → |
| hermes-workspace | ★ 6.7K | AI Agents | JavaScript | agent-ui · ai-workspace · multi-agent · swarm-orchestration · chat-ui · memory · skills · mcp · terminal · electron · self-hosted · pwa · dashboard · openai-compatible · hermes-agent · control-plane | Electron/web workspace UI for Nous Research's Hermes Agent: streaming chat, memory browser, skills and MCP management, terminal, and a multi-agent Swarm control plane. Runs zero-fork against the vanilla hermes-agent gateway plus dashboard API. | View Project → |
| LaVague | ★ 6.4K | AI Agents | Python | web agents · browser automation · large action models · LLM · RAG · web testing · automation | LaVague is a Python framework for building AI web agents that turn natural-language objectives into browser actions. It supports Selenium, Playwright, and a Chrome extension driver, with optional Gradio demos and QA test automation. | View Project → |
| deepchat | ★ 6.3K | AI Agents | TypeScript | desktop-client · local-first · electron · mcp-client · agent-skills · acp · multi-model · ollama · cross-platform · chat-ui · remote-control · session-trace · deeplink · multi-modal · web-search · typescript | DeepChat is a local-first Electron desktop client for AI agents: run cloud or Ollama models, ACP agents and MCP tools in one app, with installable Skills, session Tape/Trace, web search and remote control via Telegram, Discord or Feishu. | View Project → |
| Kun | ★ 6.3K | AI Agents | TypeScript | local-first · multi-agent · coding-agent · desktop-app · TUI · MCP · skills · automation · workspaces · computer-use | Kun is a local-first AI agent workspace with desktop GUI and TUI sharing one runtime. Agents can plan and execute coding, writing, design, research, and automation tasks, with multi-agent collaboration, MCP, skills, approvals, and evidence tracking. | View Project → |
| BrowserSkill | ★ 6.3K | AI Tools | TypeScript | browser-automation · browser-use · computer-use · cli · browser-extension · agent-skills · human-in-the-loop · chrome · edge · logged-in-session · daemon · screenshots · web-automation · agent-harness · typescript · rust | BrowserSkill is a CLI plus browser extension that lets shell-capable AI agents drive your already logged-in Chrome/Edge profile in a separate Agent Window, with explicit tab borrowing and human-in-the-loop handoff for captchas or logins. | View Project → |
| fast-jev-compaction | ★ 6.3K | AI Coding | TypeScript | context-compaction · claude-code · claude-code-plugin · context-window · token-reduction · memory-management · llm-context · session-history · verbatim-context · typescript · npm-library · hooks | Claude Code plugin and npm library that swaps the built-in compaction summary for Jev-scored pruning: every tool call and result is judged in one fast request, stale ones dropped or truncated, kept messages stay verbatim. Falls back to the built-in summary when reduction is too small or Jev fails. | View Project → |
| Everywhere | ★ 6.3K | AI Agents | C# | desktop assistant · computer-use agent · screen awareness · accessibility APIs · UI automation · MCP · sub-agents · cross-platform | A desktop AI assistant that uses screen content, accessibility data, and active-app context to respond to requests. It supports multiple LLM providers, MCP tools, sub-agents, browser and filesystem access, and is available for Windows and macOS. | View Project → |
| Skills | ★ 6.2K | Skills | HTML | agent-skills · codex · claude-code · cursor · prompt-engineering · web-design · frontend · threejs · gsap · webgl · game-development · ui-design · workflows · markdown · skill-packs · design-systems | A curated library of 123 portable agent skills (SKILL.md playbooks) for Codex, Claude Code, Cursor and other coding agents, covering web design, Three.js games, UI prompting, media assets and reusable workflows. Load the matching skill folder before asking an agent to build. | View Project → |
| AutoHedge | ★ 6.2K | AI Agents | Python | multi-agent · trading · finance · autonomous-agents · swarm-intelligence · quantitative-trading · crypto · solana · risk-management · market-analysis · llm-agents · agent-pipeline · python · hedge-fund · trade-execution · cli | AutoHedge is a Python multi-agent framework that runs an autonomous hedge fund: a Director, Quant, Risk and Execution agent pipeline analyzes markets, sizes risk, and executes trades, currently on Solana with Coinbase planned. | View Project → |
| helicone | ★ 6.2K | AI Tools | TypeScript | llm-observability · ai-gateway · llmops · monitoring · evaluation · tracing · cost-tracking · prompt-management · playground · analytics · logging · routing · fallbacks · self-hosted · clickhouse · openai-compatible | Helicone is an open source LLM observability platform and AI gateway. One line of code logs OpenAI, Anthropic, LangChain and other provider calls for tracing, cost/latency analytics, prompt management and evaluation. Self-hostable via Docker with ClickHouse and Supabase. | View Project → |
| atlas | ★ 6.2K | AI Coding | TypeScript | ai-coding · coding-agents · multi-agent · claude-code · codex · acp · agent-client-protocol · git · source-control · checkpoints · session-tracking · agent-memory · skills · mcp-client · local-first · self-hosted | Atlas is a desktop source-control layer for coding agents: it runs Claude Code, Codex, Atlas's own agent and ACP registry agents side by side, records every session locally, and links each git commit back to the prompts, tool calls and reasoning behind it. Adds shared on-device memory, skills, knowledge notes and a que | View Project → |
| qiaomu-anything-to-notebooklm | ★ 6.2K | Skills | Python | claude-skill · notebooklm · content-processing · podcast-generation · mindmap · ppt · quiz · paywall-bypass · wechat · youtube · pdf · epub · transcription · mcp · playwright · markitdown | A Claude Code Skill that pulls in content from 15+ sources (WeChat, YouTube, PDF/EPUB, podcasts, paywalled sites, search) and uploads it to Google NotebookLM to generate podcasts, PPTs, mindmaps, quizzes or deep-analysis reports via natural language. | View Project → |
| semble | ★ 6.1K | AI Tools | Python | code search · semantic search · agent tooling · MCP server · embeddings · retrieval · token-efficient | Semble indexes code locally and provides fast semantic search for coding agents through a Python library, CLI, or MCP server. It returns ranked code snippets and supports local or remote repositories without requiring external services or API keys. | View Project → |
| ouroboros | ★ 6.1K | AI Coding | Python | agent-os · ai-coding-agent · mcp-server · spec-driven-development · llm-orchestration · llm-evaluation · socratic-interview · multi-runtime · claude-code · codex-cli · agentic-loop · workflow-engine · developer-tools · replayable-runs · cli · python | Ouroboros is a local-first Agent OS for AI coding: it replaces ad-hoc prompting with a spec-first loop (interview, crystallize, execute, evaluate, evolve). Ships as an MCP server plus CLI adapters for 14 runtimes including Claude Code, Codex CLI, Gemini CLI and OpenCode. | View Project → |
| loopx | ★ 6.1K | AI Agents | Python | agent-control-plane · long-horizon-agents · multi-agent · durable-state · agent-harness · local-first · workflow-automation · task-orchestration · agent-teams · human-in-the-loop | LoopX is a local-first control plane for long-running agents and agent teams. It preserves goals, decisions, evidence, ownership, and continuation state across sessions while connected runtimes such as Codex or Claude Code perform the work. | View Project → |
| LobsterAI | ★ 6.1K | AI Agents | TypeScript | desktop-agent · computer-use · multi-agent · skills · mcp-client · office-automation · im-remote-control · scheduled-tasks · browser-automation · artifact-rendering · local-memory · electron · cross-platform · openclaw · approval-gating · self-hosted | LobsterAI is a NetEase Youdao desktop AI agent (Electron/React) built on the OpenClaw runtime that executes real work on your machine: local file and data analysis, docs/slides, browser automation, video generation, plus 28 built-in skills, MCP servers and scheduled tasks. It supports multi-agent roles, local SQLite me | View Project → |
| agents-cli | ★ 6K | AI Tools | Python | AI agents · coding assistant skills · Google Cloud · agent development · agent evaluation · deployment · CLI · observability · CI/CD | A Python CLI and skill pack that helps coding assistants build, evaluate, and deploy Google ADK agents. Includes project scaffolding, evaluation workflows, Google Cloud deployment, and Gemini Enterprise publishing. | View Project → |
| shimmy | ★ 5.9K | AI Frameworks | Rust | LLM inference · local AI · inference server · WebGPU · GGUF · OpenAI-compatible API · Rust · GPU acceleration · streaming | Shimmy is a Rust inference server for local GGUF language models, with WebGPU acceleration through Airframe and OpenAI-compatible chat, completion, and streaming APIs. It runs as a single binary without Python or llama.cpp. | View Project → |
| open-connector | ★ 5.9K | MCP | TypeScript | mcp · oauth · connectors · agent-tools · saas-integrations · api-gateway · sdk · cli · openapi · self-hosted · cloudflare-workers · typescript · credentials · action-runtime · docker | OpenConnector is a self-hostable auth gateway that links 1000+ SaaS providers and 10,000+ Actions to AI agents via MCP, SDK, CLI, HTTP, and OpenAPI. It keeps OAuth credentials behind a runtime boundary and ships a dashboard, policy controls, and run logs. | View Project → |
| nexent | ★ 5.9K | AI Agents | Python | no-code · agent-generation · multi-agent · a2a · mcp · rag · memory · orchestration · skills · knowledge-base · multimodal · multi-tenant · agent-marketplace · self-hosted · nl2agent · llm | Nexent is a self-hostable, zero-code platform that turns natural-language descriptions into runnable AI agents, bundling tools, skills, layered memory, RAG knowledge bases, MCP integration and A2A multi-agent orchestration with Docker/Kubernetes deployment. | View Project → |
| PI-Desktop | ★ 5.8K | AI Coding | TypeScript | desktop-app · coding-agent · local-first · electron · rust · mcp · plugins · skills · subagents · multi-model · i18n · session-management · agent-harness · plan-mode · permissions · openai-compatible | PI-Desktop is a local-first Electron + Rust desktop workspace for AI coding agents. Open a local project, connect any model (OpenAI, Anthropic, Ollama, OpenAI-compatible), and let agents read, edit, run commands, and iterate under a permission layer with Agent, Plan, and Goal modes plus plugins, Skills, MCP, and subage | View Project → |
| laya-mlx | ★ 5.8K | AI Frameworks | Python | mlx · apple-silicon · inference-runtime · local-inference · on-device-ai · typed-decisions · modernbert · python · fp16 · structured-output · batch-inference · model-conversion · huggingface-hub · macos · low-latency · no-cloud | Native Apple Silicon MLX inference runtime for Laya typed-decision models. Returns choice probabilities, rubric scores and P(true) locally in ~7-14 ms per short question, with no PyTorch, tokenizer decoding or cloud API at runtime. | View Project → |
| higgsfield | ★ 5.8K | AI Frameworks | Jupyter Notebook | distributed-training · gpu-orchestration · llm-training · pytorch · deepspeed · zero-3 · fsdp · fault-tolerance · cluster-management · mlops · multi-node · sharding · llama · training-framework · github-actions · checkpointing | Higgsfield is an open-source GPU orchestration and ML training framework for multi-node training of billion-to-trillion parameter LLMs. It wraps PyTorch FSDP and DeepSpeed ZeRO-3 with node allocation, experiment queuing, monitoring, and GitHub Actions-driven deployment across cloud nodes. | View Project → |
| OpenResearch | ★ 5.8K | AI Agents | Rust | research-agents · autoresearch · agent-orchestration · local-first · cli · dashboard · git-worktrees · experiment-tracking · reproducibility · multi-agent · remote-compute · skills · rust · sqlite · arxiv · ml-research | Local-first Rust CLI and dashboard (orx) that turns coding agents like Claude Code, Codex, OpenCode, and Cursor into research agents: parallel git-worktree sessions, git-native experiment tracking, and remote compute on SSH/Slurm/K8s. Stores everything in local SQLite on 127.0.0.1. | View Project → |
| lemonade | ★ 5.8K | AI Frameworks | C++ | local-llm · llm-inference · openai-api · onnxruntime · llamacpp · npu · rocm · vulkan · multimodal · gguf · text-to-speech · image-generation · mcp-server · amd · local-server · quantization | Lemonade is a local AI server that runs optimized LLMs, speech, and image models on your own GPU/NPU, exposing OpenAI, Anthropic, and Ollama compatible APIs. Ships a CLI, model manager, and MCP server for connecting desktop apps and coding agents to private on-device inference. | View Project → |
| coze-loop | ★ 5.7K | AI Tools | Go | llmops · agentops · prompt-management · prompt-versioning · evaluation · observability · tracing · playground · monitoring · ai-agents · multi-model · go · self-hosted · sdk | Coze Loop is a self-hosted LLMOps platform for AI agent teams: visual prompt playground with versioning, automated multi-dimensional evaluation of prompts and agents, and trace-based observability. Ships SDKs and supports OpenAI and Volcengine Ark models via Eino; deploy with Docker Compose or Helm. Go backend, Apache- | View Project → |
| ruoyi-ai | ★ 5.7K | AI Frameworks | Java | multi-agent · rag · mcp · workflow-orchestration · knowledge-base · llm-gateway · vector-database · supervisor-mode · visual-workflow · enterprise · self-hosted · java · spring-boot · langchain4j · docker · skills | Java/Spring Boot enterprise AI platform on Langchain4j: multi-provider LLM management, local RAG with Milvus/Weaviate/Qdrant, MCP tool and Skill integration, visual workflow orchestration, and Supervisor-mode multi-agent coordination with admin and user frontends. | View Project → |
| OpenBot | ★ 5.7K | AI Agents | TypeScript | ag-ui · agent-governance · browser-automation · computer-use · mcp · self-hosted · audit-log · policy-as-code · generative-ui · multi-agent · coworkers · docker · tauri · cel-policy · gvisor-sandbox · skills | OpenBot is CopilotKit's self-hosted platform for AI coworkers: each bot gets its own container with a browser, files and shell, and every action passes through a fail-closed policy gateway that records it before executing. Any AG-UI agent (LangGraph, CrewAI, Mastra, ADK) plugs in unchanged. | View Project → |
| maka | ★ 5.7K | AI Agents | TypeScript | agent-runtime · event-sourcing · local-first · desktop · electron · cli · tui · tool-use · computer-use · mcp · multi-agent · durable-log · sqlite · privacy · benchmarking · typescript | Apache Maka (Incubating) is a local-first agent workspace where every model message, tool call, permission decision and termination is an append-only RuntimeEvent; UI, next prompt and crash recovery are projections of that log. Desktop, TUI and CLI are thin clients of one Runtime Host, and you bring your own cloud or l | View Project → |
| lieflat-charts | ★ 5.7K | Skills | HTML | agent-skills · data-visualization · charts · html · svg · claude-code · codex · moxt · report-generation · templates · dashboards · design-system · color-presets · interactive-charts | Lieflat Charts is an Agent Skill (SKILL.md) that turns raw data into polished, interactive HTML charts and 12 full-page report templates, usable in Claude Code, Codex and Moxt agents. It bundles 49 chart templates (Lupi/Glance/Basics), SVG, Chart.js and ECharts styles plus Mono and 3 color presets. | View Project → |
| openagent | ★ 5.7K | AI Agents | Go | agent · agentic · rag · mcp · computer-use · browser-use · coding-agent · multi-agent · knowledge-base · self-hosted · workflow-automation · personal-assistant · tool-calling · embeddings · semantic-search · go | OpenAgent is a self-hosted personal AI assistant in Go that combines 30+ LLM providers, RAG knowledge bases, MCP tool integration, and autonomous agent loops for browser-use, computer-use, and coding tasks, shipped as a single binary with workflow automation and admin analytics. | View Project → |
| mcp-playwright | ★ 5.7K | MCP | TypeScript | playwright · browser-automation · mcp-server · model-context-protocol · web-scraping · screenshots · codegen · javascript-execution · device-emulation · e2e-testing · claude-desktop · cursor · vscode · http-transport · stdio · docker | MCP server that exposes Playwright browser automation to LLM clients: navigate, click, screenshot, scrape, run JS, generate test code, and emulate 143 devices. Runs over stdio or standalone HTTP, installed via npx in Claude Desktop, Cursor, Cline and VS Code. | View Project → |
| hister | ★ 5.6K | MCP | Go | search-engine · personal-search · privacy · full-text-search · semantic-search · mcp-server · browser-history · self-hosted · local-first · embeddings · indexing · crawler · browser-extension · tui · go · bookmark-import | Self-hosted personal search engine in Go: full-text indexes browsing history, local files, and bookmarks, with optional embedding-based semantic search plus an MCP server so AI assistants can query your own index. | View Project → |
| Decepticon | ★ 5.6K | AI Agents | Python | autonomous-agents · multi-agent · cybersecurity · pentesting · red-team · langgraph · langchain · llm-orchestration · kali-linux · c2 · mitre-attack · docker-sandbox · knowledge-graph · cli · benchmarks · offensive-security | Autonomous red team agent built on LangGraph/LangChain that runs full kill chains (recon, exploitation, privesc, lateral movement, C2) with 16 specialist agents inside an isolated Kali Docker sandbox. Supports tiered multi-provider LLM fallback and engagement planning (RoE, OPPLAN, MITRE ATT&CK). | View Project → |
| short-video-factory | ★ 5.5K | AI Tools | TypeScript | short-video · video-generation · auto-editing · text-to-speech · subtitles · marketing · batch-processing · electron · desktop-app · automation · content-creation · cross-platform · openai-compatible · tts · video-clipping | Cross-platform Electron desktop app that turns a prompt plus video clips into finished short videos: AI script generation, TTS voiceover, subtitle effects, and batch auto-editing via FFmpeg. Runs locally with an OpenAI-compatible endpoint and EdgeTTS. | View Project → |
| ip-as-logo-skill | ★ 5.4K | Skills | — | agent-skill · image-generation · logo-design · mascot-design · prompt-engineering · brand-identity · character-design · codex-skill · agent-skills-format · text-to-image · design-system · creative-workflow | A compact Agent Skill (SKILL.md) that guides any compatible AI agent to batch-generate ultra-simple, cute, rounded IP mascot logo concepts with strict complexity limits and fixed 3-color, lower-corner composition rules. Requires a top-tier image model; no scripts or dependencies. | View Project → |
| openmed | ★ 5.4K | AI Tools | Python | clinical NLP · medical NER · PII detection · de-identification · local AI · on-device inference · healthcare · privacy · multilingual · model catalog · MCP · agent skills | OpenMed is a local-first clinical NLP SDK for medical entity extraction and PII detection and de-identification. It supports Python and mobile/browser runtimes, with model execution paths including MLX and ONNX Runtime. | View Project → |
| ZCode | ★ 5.4K | AI Coding | TypeScript | coding-agent · agent-harness · cli · tui · electron · desktop-app · web-ui · monorepo · multi-provider · terminal-agent · code-assistant · rpc · self-hosted · typescript · remote-development · llm-runtime | ZCode is Z.ai's AI coding agent harness shipped as an Electron desktop app, a browser workbench, and a terminal agent (TUI/CLI). It bundles the agent runtime, multi-provider config, RPC layer, and packaging scripts in a TypeScript pnpm monorepo. | View Project → |
| lms | ★ 5.3K | AI Tools | TypeScript | cli · llm · local-inference · model-management · lm-studio · model-serving · openai-compatible-api · gguf · developer-tools · typescript · nodejs · llm-ops · terminal · sdk-scaffolding | Official CLI for LM Studio that manages local LLM inference: start/stop the local API server, list downloaded or loaded models, load/unload models with GPU options, stream logs, and scaffold SDK projects. JSON output modes make it scriptable for local model ops. | View Project → |
| OpenExecutive | ★ 5.2K | AI Agents | Python | multi-agent · executive assistant · business · orchestration · RAG · episodic memory · MCP · Slack · Discord · Telegram | An AI executive assistant that routes work to eight specialist agents while presenting a consistent persona. It combines company-document RAG, episodic memory, proactive scheduling, and integrations for chat platforms and MCP clients. | View Project → |
| Peekaboo | ★ 5.2K | MCP | Swift | macos · mcp · screenshot · ui-automation · screen-capture · accessibility · computer-use · ai-agents · swift · cli · vqa · menu-bar · automation · mcp-server · image-analysis · vision | Peekaboo is a macOS CLI and optional MCP server that lets AI agents capture screenshots, inspect app accessibility trees, and click or type in native UIs. It supports local or remote model-based visual Q&A and exposes its tools to MCP clients like Claude Code or Cursor. | View Project → |
| ArcReel | ★ 5.2K | AI Tools | Python | ai-video · text-to-video · image-to-video · storyboard · novel-to-video · script-to-video · agentic-workflow · self-hosted · docker · ffmpeg · jianying-draft · multi-provider · cost-tracking · character-consistency · short-drama · python | Self-hosted AI video production workspace that turns novels, scripts or product assets into consistent characters, scenes, storyboards and video clips, with agent-orchestrated stages, multi-provider models, cost tracking and CapCut/Jianying draft export. | View Project → |
| code-graph-rag | ★ 5.2K | RAG | Python | rag · knowledge-graph · code-analysis · tree-sitter · mcp-server · memgraph · monorepo · semantic-search · ast · code-editing · dead-code-detection · multi-language · cypher · codebase-search · graph-database | Code-Graph-RAG parses a multi-language monorepo with Tree-sitter, stores functions, classes and relationships in a Memgraph knowledge graph, and answers natural-language queries that generate Cypher. It also edits code via AST patches, finds dead code, merges runtime traces, and runs as an MCP server for Claude Code. | View Project → |
| go-sdk | ★ 5.2K | MCP | Go | MCP · Go · SDK · client · server · OAuth · JSON-RPC · stdio · HTTP transport | Official Go SDK for implementing MCP clients and servers. Includes protocol APIs, JSON-RPC and transport support, OAuth primitives, and examples. | View Project → |
| zotero-mcp | ★ 5.1K | MCP | Python | mcp · mcp-server · zotero · research-library · citations · semantic-search · pdf-annotations · bibliography · claude · chatgpt · cli · agent-skill · embeddings · academic-research · local-api · fastmcp | Zotero MCP is a Model Context Protocol server (plus zotero-cli and an agent skill) that connects a Zotero research library to Claude, ChatGPT, and other assistants. It supports library search, semantic search, PDF reading and annotation, and adding or organizing items. | View Project → |
| tabularis | ★ 5.1K | MCP | TypeScript | mcp-server · mcp · sql · database-gui · database-client · sql-editor · sql-notebook · desktop-app · tauri · rust · text-to-sql · er-diagram · visual-explain · plugin-system · postgresql · duckdb | Tabularis is a Tauri/React desktop SQL workspace for PostgreSQL, MySQL/MariaDB, SQLite plus 16 plugin drivers including DuckDB, ClickHouse and Redis. It ships an MCP server so Claude, Cursor and Devin can read schemas and run queries, and adds SQL notebooks, visual EXPLAIN and local Ollama text-to-SQL. | View Project → |
| Kiln | ★ 5.1K | AI Tools | Python | AI development · evaluations · prompt optimization · RAG · AI agents · fine-tuning · synthetic data · dataset management · MCP · local-first · desktop app · Python library | Kiln is a local-first desktop workbench and Python library for developing AI systems. It combines evaluations, prompt optimization, RAG, agents, fine-tuning, synthetic data generation, and dataset management. | View Project → |
| PPTAgent | ★ 5.1K | AI Agents | Python | presentation-generation · powerpoint · agentic-framework · slide-generation · mcp · agent-skills · multimodal-review · deep-research · text-to-image · pptx-export · visual-review · code-generation · reflection · document-generation | PPTAgent is an agentic framework that generates, revises, and visually reviews editable PowerPoint decks. It runs as an MCP server or as a skill for Claude Code, Codex, and OpenCode, with deep research, text-to-image assets, and multimodal slide review. | View Project → |
| agentgateway | ★ 5K | MCP | Rust | mcp · mcp-gateway · ai-gateway · llm-gateway · a2a · agents · kubernetes · gateway-api · reverse-proxy · service-mesh · guardrails · rate-limiting · observability · rbac · rust · inference-routing | Rust-based AI-native proxy that sits in front of LLM providers, MCP servers, and A2A agents, adding auth, RBAC, guardrails, rate limiting, and OpenTelemetry. Runs standalone or on Kubernetes via Gateway API, with tool federation and inference-aware routing. Apache-2.0, Linux Foundation project. | View Project → |
| exa-mcp-server | ★ 5K | MCP | TypeScript | mcp · mcp-server · web-search · web-crawling · content-fetching · model-context-protocol · deep-research · agent-skills · code-search · oauth · streamable-http · stdio · typescript · exa | Exa's official MCP server exposes web search, page fetching/crawling, and Exa Agent multi-step research to any MCP client via hosted streamable HTTP or local stdio. Ships default and optional tools plus two agent skills for deep research and enrichment. | View Project → |
| mcporter | ★ 5K | MCP | TypeScript | mcp · mcp-client · cli · typescript · tool-calling · codegen · oauth · daemon · record-replay · chrome-devtools · mcp-servers · typed-clients · json-schema · agent-skills · streamable-http · stdio | TypeScript runtime and CLI for discovering and calling MCP servers. Masquerades MCP tools as typed TS APIs, or packages them into standalone CLIs; supports OAuth, config import from other MCP clients, record/replay, and a pooled daemon. | View Project → |
| mistral-vibe | ★ 5K | AI Coding | Python | cli · coding-assistant · terminal · mistral · subagents · mcp · skills · slash-commands · tool-permissions · voice-mode · multimodal · opentelemetry · python · developer-tools | Mistral Vibe is Mistral's open-source CLI coding agent in Python. It offers an interactive terminal chat with file edit, shell, grep and git tools, built-in agent profiles, subagents, skills, MCP servers and ACP editor integration. | View Project → |
| mockserver-monorepo | ★ 5K | AI Tools | Java | mock-server · api-testing · proxy · http2 · grpc · websockets · chaos-engineering · openapi · mcp · llm-mocking · json-rpc · kafka · mqtt · docker · kubernetes · tcp | MockServer is a Java HTTP(S) mock server and proxy for testing: mock APIs, record/modify live traffic, inject failures, and cover HTTP/1.1, HTTP/2, gRPC, WebSockets, TCP and brokers. It also mocks LLM chat APIs and ships a built-in MCP server, but is primarily a general API testing tool. | View Project → |
| ag2 | ★ 5K | AI Agents | Python | multi-agent · agent-framework · autogen · agentic-ai · llm · python · mcp · a2a · human-in-the-loop · tool-calling · orchestration · async · agentos · genai · network-orchestration · structured-outputs | AG2 (formerly AutoGen) is a Python framework and async AgentOS for building LLM agents and orchestrating multi-agent systems over a Network of hubs and channels, with tools, human-in-the-loop hooks, structured outputs and telemetry. | View Project → |
| ai-agents-from-zero | ★ 5K | AI Agents | Python | AI agents · tutorial · multi-agent · RAG · MCP · agent skills · LLM applications · fine-tuning · enterprise deployment · interview preparation | A Chinese-language, hands-on guide to building AI agent applications, covering LangChain, LangGraph, MCP, RAG, fine-tuning, deployment, and interview preparation. Includes tutorials and projects for multi-agent research and natural-language data querying. | View Project → |
| open-science | ★ 4.9K | AI Agents | TypeScript | scientific-agents · local-first · model-agnostic · reproducible-research · mcp · electron · desktop-app · research-workbench · python · r · notebooks · data-connectors · provenance · bioinformatics · human-in-the-loop · skills | Open-source, local-first desktop AI research workbench where scientific agents combine MCP tools, permissioned data connectors, and Python/R notebooks to produce reproducible, provenance-tracked reports, tables, and figures. Model-agnostic across providers; ships for macOS, Windows, and Linux. | View Project → |
| zep | ★ 4.9K | AI Agents | Python | agent-memory · knowledge-graph · temporal-knowledge-graph · llm-memory · integrations · ingestion · benchmarks · evals · python · typescript · go · crewai · langgraph · autogen · pydantic-ai · mcp-config | Zep's examples, integrations, and tools repo for building agent memory on Zep Cloud. Includes framework integration packages (LangGraph, CrewAI, AutoGen, Google ADK, Mastra), a bulk ingestion pipeline, memory benchmarks, and an eval harness. Note: this is not Zep's product code; the open-source temporal graph engine li | View Project → |
| ha-mcp | ★ 4.9K | MCP | Python | home-assistant · smart-home · home-automation · mcp-server · model-context-protocol · natural-language-control · fastmcp · hacs · docker · add-on · oauth · oidc · webhook · automation · iot · python | Unofficial Home Assistant MCP server exposing 87+ tools so AI assistants like Claude and ChatGPT can control devices, query states, call services and edit automations. Runs in-process via HACS, as an HA add-on, Docker, or uvx/stdio. | View Project → |
| agenta | ★ 4.8K | AI Agents | TypeScript | agent workspace · agent builder · agent orchestration · background agents · agent observability · MCP · skills · team collaboration · workflow automation · self-hosted | Agenta is a self-hostable workspace for building, running, and sharing AI agents. It supports multiple agent harnesses and models, MCP and app integrations, scheduled or event-triggered runs, permissions, and execution tracing. | View Project → |
| agent-device | ★ 4.8K | MCP | TypeScript | mcp · mcp-server · mobile-automation · ios · android · harmonyos · react-native · flutter · e2e-testing · ai-agents · coding-agents · accessibility-snapshots · screenshot-evidence · ci-cd · adb · device-cloud | CLI, stdio MCP server, and typed Node.js API that give AI coding agents a live feedback loop: they inspect, drive, and verify iOS, Android, HarmonyOS, TV, web, macOS, and Linux apps via accessibility snapshots, then save screenshots, logs, and replay scripts for CI evidence. | View Project → |
| tradingview-mcp | ★ 4.7K | MCP | Python | market data · technical analysis · backtesting · stock screener · cryptocurrency · forex · futures · sentiment analysis · TradingView · MCP server · walk-forward testing | A Python MCP server that gives AI clients access to market data, screeners, technical analysis, financial news and sentiment, and strategy backtesting. Supports stocks, crypto, forex, and futures; can be self-hosted or used through a hosted endpoint. | View Project → |
| obsidian-mind | ★ 4.7K | AI Agents | TypeScript | persistent memory · AI coding agents · Obsidian · knowledge management · agent hooks · semantic search · second brain · multi-agent support | An Obsidian vault template that gives Claude Code, Codex CLI, and Gemini CLI persistent, git-tracked memory. Hooks and agent commands manage session context and notes, with optional QMD semantic search and MCP access. | View Project → |
| claude-code-guide | ★ 4.6K | AI Coding | Python | claude-code · documentation · cheat-sheet · cli · slash-commands · mcp · sub-agents · skills · hooks · configuration · workflows · automation · anthropic · troubleshooting · vscode · best-practices | Community-maintained reference guide for Anthropic's Claude Code CLI: install, slash commands, configuration, MCP integration, sub-agents, skills, hooks, security and troubleshooting. Best as a lookup cheatsheet rather than runnable code. | View Project → |
| Claude-BugHunter | ★ 4.6K | Skills | Python | claude-code · agent-skills · bug-bounty · pentesting · red-team · offensive-security · recon · osint · vulnerability-research · web-security · api-security · triage · reporting · cloud-security · burp-mcp · security-tools | Claude Code Agent Skills bundle (83 skills, 15 slash commands, 681 disclosed-report patterns) that turns Claude into a bug-bounty hunter and external red-team operator: web/API vuln classes, enterprise CVE chains, recon, triage and reporting, with scope gates. | View Project → |
| 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 → |
Vibe Coding Discover