MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Open-source, self-hostable Zendesk and Intercom alternative for AI customer support: chat widget, shared inbox, real-time voice, Multimodal RAG, and MCP.
npm MIT ★ 22 -
Self-hosted SDLC context graph + RAG over GitLab & GitHub repos: code graph (definitions, imports), MR diffs, CI, planning, security — served to AI agents via MCP. A free, lightweight OSS alternative to GitLab Orbit (Qdrant + NATS; no ClickHouse/K8s).
MIT -
A lightweight MCP server for local documentation - simple alternative to RAG pipelines
MIT -
Offline AI meeting notes and transcription for Linux and macOS. Open-source Google AI Edge Foresight and Granola alternative with local Gemma LLM, Whisper, RAG and MCP.
MIT ★ 4 -
swarmvaultgithub.com/swarmclawai/swarmvaultupdated this year
The local-first LLM Wiki: open-source knowledge graph builder, RAG knowledge base, and agent memory store. Built on Andrej Karpathy's pattern. An Obsidian alternative for personal knowledge management, AI second brain, and durable Claude Code / Codex / OpenClaw memory.
npm MIT ★ 736 -
Website to Clean Markdown, Vector RAG Chunks & llms.txt ($0.50/1k) — Includes 1-Click n8n Workflow, Make.com Blueprint & Claude MCP Server (https://apify.com/lucacelli/ai-ready-website-to-markdown-rag-crawler)
-
flux-memorygithub.com/harsh5i/flux-memoryupdated this year
Self-organizing retrieval fabric for AI memory — deployable MCP server, REST API, and Python SDK
MIT ★ 4 -
Web browsing superpowers for AI agents - one MCP server to scrape, crawl, extract, map, search, batch, research & monitor the web. Self-hosted, no API keys. Free Firecrawl alternative.
pypi MIT ★ 25 -
Institutional-grade AI pipeline that generates manager tearsheets for hedge funds, PE firms, and family offices. MCP-enabled.
-
llm-wikimindgithub.com/HAL-9909/llm-wikimindupdated this year
The production-ready implementation of Karpathy's LLM Wiki pattern. Markdown + BM25 + MCP. No embeddings, no vector DB.
MIT ★ 2 -
Context.dev for OpenClaw: live web search, scraping and structured extraction for agents.
MIT ★ 2 -
wikiforgegithub.com/sawzhang/wikiforgeupdated this year
Self-hosting LLM knowledge platform: scrape sources → compile into a living wiki → serve to humans (web+chat), agents (MCP), and programs (API). Compile, don't retrieve.
MIT -
Context.dev plugin for Cursor: live web search, scraping and structured extraction over MCP.
MIT ★ 11 -
SpecRouterMCPgithub.com/prabhukirangit/SpecRouterMCPupdated this year
SpecRouterMCP is a high-performance Model Context Protocol (MCP) server that dynamically converts massive OpenAPI/Swagger specifications into executable LLM tools on-the-fly.
Apache-2.0 -
Context.dev extension for Gemini CLI: live web search, scraping and document parsing.
MIT -
Repository-scoped memory for coding agent's decisions, pitfalls, and handoffs stored in one inspectable SQLite file, so agents continue instead of starting over.
NOASSERTION ★ 24 -
unsearchgithub.com/Rakesh1002/unsearchupdated this year
Open-source web search API for AI agents. Tavily-compatible. Apache 2.0, self-hostable on Cloudflare Workers. 10× cheaper.
Apache-2.0 ★ 9 -
The open-source context layer for AI agents and enterprise AI search. Permission-aware search and cited answers across 30+ sources, including Slack, Drive, Jira, GitHub and Microsoft 365, for people and agents through MCP, SDKs and built-in agents. Self-hosted or Cloud.
go Apache-2.0 ★ 4,123 -
firescraplinggithub.com/grayaa/firescraplingupdated this year
Self-hosted cost-control layer for web scraping. Bring your own Scrape.do/Scrapfly key, escalate fetch tiers only when needed, get LLM-ready Markdown. MCP server included.
NOASSERTION -
URL to clean Markdown for AI agents. 2 tools: scrape_url (one page) and crawl_website (whole site). Blocked or paywalled pages are reported honestly, never invented. Hosted endpoint at scrape.mcp.goaimoat.com/mcp, free 3 calls per email, no card. A cheaper Firecrawl alternative; also runs locally via Docker.
MIT -
Self-hosted MCP server and Tavily-compatible API for AI agents: web search, extract, crawl and map in one lightweight CPU-only Python process. JS pages via headless Chromium, PDF, no API keys.
AGPL-3.0 ★ 3 -
Memory that forgets: a persistent, self-revising belief-state MCP server and Python API for agentic R&D. Uses a hypothesis DAG to auto-prune dead ends and deduce truth without extra probes.
pypi MIT ★ 13 -
fathom-mcpgithub.com/RomanShnurov/fathom-mcpupdated this year
An MCP server for local file knowledge. Bridges file systems and LLMs using ugrep and poppler-utils to offer full-text search, fuzzy matching, and hierarchical context access for PDFs and codebases, all in a read-only, secure environment.
pypi MIT -
给 AI Agent 的长期记忆库:会遗忘、会弃答、零依赖 · Long-term memory for LLM agents with forgetting curve, abstention, SQLite/WAL, multi-process writes and an MCP server. No vector DB required.
MIT ★ 1 -
hcegithub.com/nakurian/hceupdated this year
Middleware memory engine for LLMs — replaces linear chat history with three parallel structures (Entity Graph, Semantic Tree, Focus Buffer) to retrieve only the most relevant context within a token budget. Works as a Python library or MCP server for Claude Code, Copilot CLI, and other AI tools.
★ 2 -
DSL based REST service for your Sqlite database (and not only) with LUA support, consider it as an alternative to PostgREST
MIT ★ 10 -
Open-source, self-hosted Tavily alternative: web search, extract, crawl and map API for AI agents, with an MCP server
pypi MIT ★ 3 -
Convert PDF, Word, PowerPoint, HTML, email & 40+ formats to clean Markdown — and back. Built for LLMs, RAG & Python pipelines, with a built-in MCP server.
pypi MIT ★ 35 -
Lightweight QMD alternative: fully local hybrid search engine (BM25 + vector + rerank) and MCP server in a single ~15MB Go binary. No Node.js. Runs on Raspberry Pi.
go MIT ★ 7 -
Eight-platform full-stack AI operating system - unifies 176 LLMs via LangGraph + MCP + A2A. Multi-tenant RLS over 340 tables, RAG knowledge base, agent marketplace. Web/API/CLI/Desktop/Extension/Mobile/Miniapp. Apache 2.0.
Apache-2.0 ★ 32 -
Open-source, self-hostable alternative to ChatGPT and NotebookLM — agents, skills, MCP, and cited knowledge in one AI workstation. Your models, your infrastructure. Includes public MCP server and agent skills directories.
npm NOASSERTION ★ 492 -
🔒 Privacy-first MCP server for Claude using PostgreSQL + Ollama. Local alternative to cloud-based code context with full data sovereignty. No API keys, no external calls, 100% local.
npm MIT ★ 16 -
Open-source metasearch backend, MCP server, and AI search API for LLM agents. Python FastAPI search gateway with Google search via SerpBase and Serper, multi-engine search aggregation, structured JSON output, provider fallback, deduplication, and SearXNG alternative architecture for agent workflows.
MIT ★ 58