Atlas Docs MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Atlas Docs MCP Server check the documentation for Astro"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Atlas Docs MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with documentation for libraries and frameworks.
Atlas Docs is currently in beta. Not everything might work perfectly, but we're actively improving the service. Your patience andfeedback are greatly appreciated!
What Does This Server Do?
LLMs are great at generating general code, but suck at correctly using less popular or newly released libraries. This isn't surprising, since the models have not been trained comprehensively on code using these libraries.
Atlas Docs MCP server:
Provides technical documentation for libraries and frameworks
Processes the official docs into a clean markdown version for LLM consumption
Is easy to set up with Cursor, Cline, Windsurf and any other MCP-compatible LLM clients
Claude 3.5 Sonnet on its own:
Claude 3.5 Sonnet with Atlas Docs MCP:
Related MCP server: Context7 MCP
📦 Installation
Atlas Docs MCP server works with any MCP client that supports the stdio protocol, including:
Cursor
Cline
Windsurf
Claude Desktop
Add the following to your MCP client configuration file:
{
"mcpServers": {
"atlas-docs": {
"command": "npx",
"args": ["-y", "@cartographai/atlas-docs-mcp"]
}
}
}That's it! You may need to restart the app (for Claude Desktop) for the server to be recognised.
Tip: Prompt your model to check the docs eg. "Use the tools to check the documentation for Astro to ensure that you use the library correctly."
Installing via Smithery
Alternatively, you can install Atlas Docs MCP automatically via Smithery. Example for claude desktop:
npx -y @smithery/cli install @CartographAI/atlas-docs-mcp --client claudeChange "claude" to "cursor", "cline" or "windsurf" for the respective clients.
📒 Available Libraries
AI-SDK (source: https://sdk.vercel.ai/docs/introduction)
Astro (source: https://docs.astro.build/en/getting-started)
ast-grep (source: https://ast-grep.github.io/llms.txt)
Bun (source: https://bun.sh/llms.txt)
CrewAI (source: https://docs.crewai.com/llms.txt)
Drizzle (source: https://orm.drizzle.team/llms.txt)
ElevenLabs (source: https://elevenlabs.io/docs/llms.txt)
Fireworks (source: https://docs.fireworks.ai/llms.txt)
Hono (source: https://hono.dev/llms.txt)
Langgraph-js (source: https://langchain-ai.github.io/langgraphjs/llms.txt)
Langgraph-py (source: https://langchain-ai.github.io/langgraph/llms.txt)
Mastra (source: https://mastra.ai/llms.txt)
ModelContextProtocol (source: https://modelcontextprotocol.io/llms.txt)
Pglite (source: https://pglite.dev/docs/about)
Prisma (source: https://www.prisma.io/docs/llms.txt)
Resend (source: https://resend.com/docs/llms.txt)
shadcn/ui (source: https://ui.shadcn.com/docs)
Stripe (source: https://docs.stripe.com/llms.txt)
Svelte (source: https://svelte.dev/docs/svelte/overview)
SvelteKit (source: https://svelte.dev/docs/kit/introduction)
tailwindcss (source: https://tailwindcss.com/docs/installation)
TanStack-Router (source: https://tanstack.com/router/latest/docs/framework/react/overview)
Trigger.dev (source: https://trigger.dev/docs/llms.txt)
X (source: https://docs.x.com/llms.txt)
Zapier (source: https://docs.zapier.com/llms.txt)
Want docs for another library not in this list? Please open an issue in this repo, we'll try to process and add it!
🔨 Available Tools
list_docs: List all available documentation setsget_docs_index: Retrieves a condensed, LLM-friendly index of a documentation setget_docs_full: Retrieves a complete documentation set in a single consolidated filesearch_docs: Search a documentation set by keywordsget_docs_page: Retrieves a specific page of a documentation set
💠How It Works
Atlas Docs processes tech libraries' documentation sites into clean, markdown versions. This MCP server provides the docs as MCP tools, calling Atlas Docs APIs for the data.
Running the backend locally
Please visit CartographAI/atlas and follow the instructions in the README. Update ATLAS_API_URL with the url of your deployment.
Support & Feedback
Please open an issue in this repo to request docs for a library, or to report a bug.
If you have any questions, feedback, or just want to say hi, we'd love to hear from you. You can find us on Cartograph's Discord comunity for real-time support, or email us at contact@cartograph.app
Available Tools
5 toolsget_docs_fullA
Retrieves the complete documentation content in a single consolidated file. Use this when you need comprehensive knowledge or need to analyze the full documentation context. Returns a large volume of text - consider using get_docs_page or search_docs for targeted information.
Inputs: docName (required).
get_docs_indexA
Retrieves a condensed, LLM-friendly index of the pages in a documentation set. Use this for initial exploration to understand what's covered and identify relevant pages. Returns a markdown page with a list of available pages. Follow up with get_docs_page to get full content.
Inputs: docName (required).
get_docs_pageA
Retrieves a specific documentation page's content using its relative path. Use this to get detailed information about a known topic, after identifying the relevant page through get_docs_index or search_docs. Returns the complete content of a single documentation page.
Inputs: docName (required), pagePath (required).
list_docsA
Lists all available documentation libraries and frameworks. Use this as your first step to discover available documentation sets. Returns name, description and source url for each documentation set. Required before using other documentation tools since you need the docName.
Inputs: no named parameters.
search_docsA
Searches a documentation set for specific content. Use this to find pages containing particular keywords, concepts, or topics. Returns matching pages ranked by relevance with their paths and descriptions. Follow up with get_docs_page to get full content.
Inputs: query (required), docName (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool has a clearly distinct purpose: listing available doc sets, searching within a set, getting an index, fetching a specific page, and retrieving full content. Even similar-sounding tools like get_docs_index and list_docs are easily differentiated by their descriptions and scope.
All tool names follow a clear verb_noun pattern using the same object prefix 'docs'. The verbs list, search, and get accurately describe the action, and the specific suffixes index, page, and full make the target resource clear.
Five tools is a well-scoped set for a documentation retrieval server. Each tool covers a necessary step in the workflow without unnecessary duplication or bloat.
The tool surface covers the complete documentation access workflow: discover doc sets, search and browse indexes, retrieve individual pages, and get the full corpus. No obvious gaps or dead ends are present.
Maintenance
Related MCP Connectors
Provide your AI coding tools with token-efficient access to up-to-date technical documentation for…
@latest documentation and code examples to 9000+ libraries for LLMs and AI code editors in a singl…
Versioned documentation registry and semantic search for AI tools and coding assistants.
Docsy gives AI assistants current, version-specific documentation for libraries, frameworks and SDKs. Resolve library IDs and versions, then search and fetch focused official documentation with original source links using five read-only MCP tools. Hosted Streamable HTTP with OAuth 2.1 sign-in.
Related MCP Servers
- AlicenseCqualityDmaintenanceProvides real-time access to up-to-date library documentation and code examples for any programming library. Helps AI coding assistants deliver accurate, current information instead of relying on outdated training data.110 npmMIT
- AlicenseAqualityDmaintenanceProvides up-to-date, version-specific documentation and code examples for libraries and frameworks directly into AI prompts, eliminating outdated code generation and hallucinated APIs.2MIT
- AlicenseAqualityDmaintenanceProvides LLMs with up-to-date, version-specific documentation and code examples directly from library sources, eliminating outdated training data and hallucinated APIs by fetching current documentation at prompt time.2MIT
- FlicenseNot gradedqualityDmaintenanceProvides real-time, up-to-date documentation for major LLM providers (OpenAI, Anthropic, Google Gemini) to prevent hallucinations and outdated code patterns in AI agents.7-