Provides tools to search and retrieve documentation from Cloudflare AI Search (AutoRAG) instances, enabling AI agents to query documentation sets stored in Cloudflare R2 buckets.
Click on "Install 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., "@docs-to-mcpSearch the documentation for how to configure the API."
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.
docs-to-mcp
A Cloudflare Worker that turns any AI Search (AutoRAG) instance into an MCP server.
Setup
Create an AI Search instance in the Cloudflare dashboard
Create a R2 bucket and upload all your docs files - markdown, pdfs, text files etc.
Go to Computer & AI > AI Search(AutorRAG)
Create a new AI Search instance and select the R2 bucket as source (the one you previously created), just select the defaults.
Configure the Worker
Update
wrangler.jsoncwith your AI Search name:"vars": { "AI_SEARCH_NAME": "your-ai-search-name" }Install dependencies
pnpm installDeploy
pnpm run deploy
Usage
MCP Endpoints
Endpoint | Transport | Description |
| Streamable HTTP | Standard MCP HTTP transport |
How did I test this.
I trained it on a fictional TypeScript framework with APIs named after my cat, Ustaad, so I would know it's not hallucinating, you can find these files in fake-knowledge.md
Available Tools
search_docs
Search the documentation and return relevant chunks.
Parameters:
query(required): The search querymax_results(optional): Maximum results (1-50, default 10)rewrite_query(optional): Rewrite query for better retrievalscore_threshold(optional): Minimum score threshold (0-1)
ai_search_docs
Search documentation and get an AI-generated response with sources.
Parameters: Same as search_docs
Connect from MCP Clients
Claude Desktop / Cursor / etc:
Direct remote MCP (if client supports it):