LlamaIndex Docs MCP lets AI agents search and navigate LlamaIndex documentation using semantic search, exact text and regex matching, documentation browsing, and direct page reading for accurate technical guidance.
Encrypted at rest, isolated from the model
Resolved from an AES-256-GCM vault at the moment of the call and attached to the request — the model never sees the secrets.
Try asking
Search through llama-index documentation using BM25 ranked search. Best for conceptual queries, finding documentation by topic, or when you're not sure of exact terminology. Returns ranked results with excerpts showing match context. Use this when: exploring unfamiliar topics, looking for 'how to' guides, searching by concept rather than exact terms, or need to understand what documentation exists on a subject. For exact term matching or code patterns, use grep_docs instead.
Search for exact text or regex patterns across documentation files. Returns all matching lines with their file locations and optional surrounding context. Best for: finding specific class/function names, locating exact error messages or code snippets, discovering all occurrences of a term, or pattern matching with regex. Unlike search_docs which ranks by relevance, this returns ALL matches in order. Use when you know precise terminology or need exhaustive results. Can return up to 500 matches vs search_docs' 20 result limit.
Read content from a specific documentation file by its path. By default returns the first 500 lines to avoid overwhelming context. Use startLine/endLine to read specific portions of longer documents. Accepts URL-style paths (e.g., '/python/framework/query-engine') or paths without extensions. Best for: reading documentation pages, understanding context after finding relevant excerpts, or accessing files you already know the path to.
Browse the documentation site structure to discover what pages and sections exist. Returns a tree showing directory slugs (usable as read_doc paths) and labels. Default depth is 1 (compact overview) when no section is specified, or 2 when a section/path is specified. Use depth=2-3 to see more detail. The path parameter performs a strict lookup — on a miss no tree is returned, just a short not-found message listing the searched sections; call again without path to browse.
One endpoint, the same key, whichever client you use.
~/Library/Application Support/Claude/claude_desktop_config.json (Mac) · %APPDATA%\Claude\claude_desktop_config.json (Windows)
Replace API_KEY with your own key.
Already have an "mcpServers" section in your config? Just add the server entry inside it.
Discovery, routing, credentials, tool scoping and execution logs all happen at the gateway→connections stay ACTIVE with no work from you
Llamaindex Docs MCP runs through a gateway that holds the credentials, scopes the access and records every call.
Managed auth, hosted MCP servers, and every Gmail tool your agent needs.
Free to start.