Test out the MCP server and toolkit in a temporary cloud Room using Meshagent. When the Meshagent Room closes (by inactivity or manually), the associated MCP server tool and chatbot are automatically removed. Required Environment Variables:
  • ATLAS_API_URL — The HTTP endpoint that the MCP server will use to source documentation.
    • Default: https://atlas.cartograph.app/api

​
Steps

  1. Install Meshagent CLI and dependencies: 
    pip install "meshagent[all]"
  2. Sign up and Authenticate:
    Follow setup steps at Meshagent CLI Getting Started.
  3. Run the Atlas Docs MCP agent inside a test Room: 
    meshagent service test \
  --room=test \
  --role=agent \
  --image=meshagent/mcp-atlas-docs:latest \
  --env MESHAGENT_PORT=8001 \
  --env ATLAS_API_URL=<YOUR_ATLAS_API_URL> \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-atlas-docs-service-test
    • This starts an MCP agent (server) in a Meshagent Room called test.
    • Replace <YOUR_ATLAS_API_URL> as needed (default is already set).
    • When the Room closes, the services are shut down.
  4. Join with a chatbot in the same Room (enabling LLM tool access): 
    meshagent chatbot join \
  --room=test \
  --agent-name=mcp-atlas-docs \
  --name=mcp-atlas-docs \
  --toolkit=mcp-atlas-docs
    • The chatbot is now in the Room and can use MCP tools via toolkit!
    • Output will provide a link to open/join the Room.
  5. Test interactive access in your browser:
    Open the provided Room link, send a message to the agent, and interact with the Atlas Docs MCP server tools through conversation.
Environment Variables (with example values)
  • ATLAS_API_URL: URL of the Atlas Docs backend endpoint (default: https://atlas.cartograph.app/api)
  • MESHAGENT_PORT: Internal port for the service (typically 8001 for Atlas Docs MCP agent, 9001 for the chatbot service)

​
Project Level Deployment (Persistent Services)

For production and continuous Room coverage:
Deploy MCP server and chatbot as persistent project-level Meshagent services. This ensures that every Room created in your Meshagent project automatically has the MCP tools and chatbot available—no manual intervention needed after setup!

​
Steps

  1. Create persistent MCP Atlas Docs agent: 
    meshagent service create \
  --role=agent \
  --image=meshagent/mcp-atlas-docs:latest \
  --env MESHAGENT_PORT=8001 \
  --env ATLAS_API_URL=<YOUR_ATLAS_API_URL> \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-atlas-docs-service
  2. Create persistent chatbot service (tool access always available): 
    meshagent service create \
  --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-atlas-docs-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-atlas-docs-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-atlas-docs-chatbot --toolkit=mcp-atlas-docs"
  • Now, every new project Room gets the MCP agent and chatbot automatically!
  • No need to run agents or chatbots on your machine — they are managed by Meshagent.

​
Tools Available

​
Atlas Docs MCP Server

The Atlas Docs MCP Server provides LLMs with hosted, clean markdown documentation of libraries and frameworks.

​
Tools Provided

ToolShort Description
get_docs_fullRetrieves the complete documentation content in a single consolidated file.
get_docs_indexRetrieves a condensed, LLM-friendly index of the pages in a documentation set.
get_docs_pageRetrieves a specific documentation page’s content using its relative path.
list_docsLists all available documentation libraries and frameworks.
search_docsSearches a documentation set for specific content.

​
Tools Details

​
get_docs_full

Retrieve the complete documentation content in a single consolidated file. Use this for full context.
Parameters:
  • docName (string): Name of the documentation set

​
get_docs_index

Retrieve a condensed, LLM-friendly index of the pages in a doc set. Great for initial exploration.
Parameters:
  • docName (string): Name of the documentation set

​
get_docs_page

Get the complete content of a specific documentation page by its path.
Parameters:
  • docName (string): Name of the documentation set
  • pagePath (string): Root-relative documentation page path (e.g., /guides/getting-started)

​
list_docs

List all available documentation libraries and frameworks — name, description, and source URL for each.

​
search_docs

Search for specific keywords or topics in a documentation set. Returns ranked matching pages.
Parameters:
  • docName (string): Name of the documentation set
  • query (string): Search query for the documentation set
For more information, see the Atlas Docs MCP Server repository and Meshagent documentation.