#5555th of 111 in MCP servers for developers

docs-mcp-server

Version-aware docs index that feeds library documentation to coding assistants

Stars
1.9k
License
MIT
Last commit
Nov 2026
Last release
Oct 2026
Language
TypeScript
Stars, last 29 weeks+634
21 Apr 2026: 1,235 stars28 Apr 2026: 1,261 stars5 May 2026: 1,286 stars12 May 2026: 1,305 stars19 May 2026: 1,323 stars26 May 2026: 1,357 stars2 Jun 2026: 1,383 stars9 Jun 2026: 1,413 stars16 Jun 2026: 1,435 stars23 Jun 2026: 1,464 stars30 Jun 2026: 1,484 stars7 Jul 2026: 1,508 stars14 Jul 2026: 1,528 stars21 Jul 2026: 1,543 stars28 Jul 2026: 1,576 stars4 Aug 2026: 1,600 stars11 Aug 2026: 1,627 stars18 Aug 2026: 1,652 stars25 Aug 2026: 1,681 stars1 Sept 2026: 1,693 stars8 Sept 2026: 1,717 stars15 Sept 2026: 1,725 stars22 Sept 2026: 1,749 stars29 Sept 2026: 1,770 stars6 Oct 2026: 1,784 stars13 Oct 2026: 1,796 stars20 Oct 2026: 1,818 stars27 Oct 2026: 1,834 stars3 Nov 2026: 1,854 stars10 Nov 2026: 1,869 stars

1,235 on 21 Apr 20261,869 on 10 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Fetches documentation from websites, GitHub repositories, npm, PyPI, local folders and archives, and indexes it per library version in a local SQLite database. Search is exposed over MCP (stdio or Streamable HTTP), a CLI and a web UI. Full-text search works by default; vector search is optional via OpenAI, Ollama, Gemini, Vertex AI, AWS Bedrock, Azure OpenAI or OpenAI-compatible servers.

Who it is for: Developers who want coding assistants to use version-matched docs

Strengths

  • Stores each library version separately, so searches match the version a project uses
  • Indexes websites, GitHub, npm, PyPI, ZIP/TAR, PDF, Office, EPUB and notebooks
  • Runs via npx or Docker with a local SQLite index, no external database
  • Works without an embedding model; Ollama keeps document text on the machine

Weaknesses

  • Usage telemetry is on by default and must be turned off manually
  • Docs indexed before an embedding model is set get no vectors until re-scraped
  • You add and refresh each library yourself; scrape refuses existing versions
  • MCP endpoint is unauthenticated unless OAuth2/OIDC is configured separately

What it needs

  • no GPU
  • Docker + Compose
  • Needs Node.js 22+ or Docker, Optional embedding provider (OpenAI, Ollama, Gemini, Vertex AI, AWS Bedrock, Azure OpenAI)
  • Models: OpenAI, Ollama, Gemini, Vertex AI, AWS Bedrock
  • port 6280

Quick answers

Is docs-mcp-server open source?

Yes. Its code is on GitHub under the MIT license.

Can you run docs-mcp-server with Docker?

Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.

Also in MCP servers for developers

See all 111
Also in MCP servers for developers
RankProjectScore
1gitnexusCode knowledge graph exposed to coding agents through MCP tools Live demo ↗ (opens in a new tab)48.4k stars+765 (+1.6%) in 30 dayscustom license96 out of 100
2codegraphLocal code knowledge graph that coding agents query over MCP77.4k stars+3,831 (+5.2%) in 30 daysMIT95 out of 100
3playwright-mcpMCP server that lets LLMs drive a browser through Playwright accessibility snapshots39.5k starsApache-2.092 out of 100
#4code-review-graphLocal Tree-sitter code graph that gives coding assistants review context over MCP32.7k stars+761 (+2.4%) in 30 daysMIT92 out of 100
#5serenaMCP server giving coding agents symbol-level code search, editing and refactoring31.3k stars+1,188 (+3.9%) in 30 dayscustom license91 out of 100