RAG MCP servers for retrieval-augmented generation
MCP servers for retrieval-augmented generation: index documents, embed chunks, and retrieve grounded context for Claude, Cursor, and other agents.
What to know about RAG MCP servers
Retrieval-augmented generation means fetching relevant passages from your own data and adding them to the prompt before the model answers. MCP turns that into a tool call: instead of wiring a retriever into every app, you run one RAG server and any compatible client can ask it for context.
RAG servers in the catalog range from complete pipelines that ingest files, chunk them, embed them, and answer queries, to thin adapters over a hosted retrieval API or a single vector store. Complete pipelines are quickest to try. Adapters suit teams that already run ingestion elsewhere and only need the query side exposed to agents.
Compare servers on three things: which embedding model they use and whether you can change it, how they chunk documents, and whether results come back with source references the model can cite. Citations are what separate a useful RAG setup from one that quietly hallucinates around your documents.
RAG MCP servers
The 12 strongest of 3493 matching servers. Servers that name the topic directly come first, then by trust score.
- 77Trust score 77 of 100, Good
Code-Graph-RAG
by vitali87 · Python
Knowledge-graph RAG for multi-language codebases: index, query and edit code in natural language.
RegistryLicenseQualityMaintenanceLicense: A. MIT is a permissive open source license.Quality: Not graded. No README or tool list has been captured yet.Maintenance: A. Last commit or release 22 days ago. - 74Trust score 74 of 100, Good
Local-first agentic RAG with citations - hybrid search, reranking, multimodal document retrieval
RegistryLicenseQualityMaintenanceLicense: A. MIT is a permissive open source license.Quality: Not graded. No README or tool list has been captured yet.Maintenance: A. Last commit or release 22 days ago. - 69Trust score 69 of 100, Good
mcp-server
by bingowon · TypeScript
Apple developer docs and WWDC transcripts with RAG retrieval and Jev relevance ranking.
RegistryRemoteLicenseQualityMaintenanceLicense: A. MIT is a permissive open source license.Quality: Not graded. No README or tool list has been captured yet.Maintenance: A. Last commit or release 12 days ago. - 60Trust score 60 of 100, Good
tw-legal-rag
by aa0101181514 · Python
Taiwan law MCP server: 22M court judgments, statutes, agency rulings, with citation checks
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: A. Last commit or release 11 days ago. - 58Trust score 58 of 100, Fair
mcp-local-rag
by shinpr · TypeScript
Easy-to-setup local RAG server with minimal configuration
RegistryLicenseQualityMaintenanceLicense: A. MIT is a permissive open source license.Quality: Not graded. No README or tool list has been captured yet.Maintenance: A. Last commit or release 15 days ago. - 55Trust score 55 of 100, Fair
Search a folder of documents and check answers against them. Offline: no API key or database.
RegistryLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. - 55Trust score 55 of 100, Fair
Author rules from policy docs, then decide: a Rete engine gives the verdict, an LLM explains why.
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. - 52Trust score 52 of 100, Fair
Transcribes audio and video files to text, SRT, VTT and timestamped chunks for RAG.
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. - 52Trust score 52 of 100, Fair
Errandkit PDF to Markdown, Tables & RAG Chunks
Unknown author
PDF, DOCX and scans to Markdown, CSV tables and RAG chunks with OCR. Runs on Apify.
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. - 52Trust score 52 of 100, Fair
Semantic boundary preserving text chunker maintaining markdown headers, code fences, and paragraphs
RegistryLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. - 52Trust score 52 of 100, Fair
LLM & RAG evals: hallucination, eval contracts, semantic tests, tokens & model info. Free, no key.
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded. 
52Trust score 52 of 100, FairStrips 85-97% of HTML/DOM boilerplate into clean Markdown, skeletons, or BM25-ranked RAG chunks.
RegistryRemoteLicenseQualityMaintenanceLicense: Not graded. No license is recorded for this server.Quality: Not graded. No README or tool list has been captured yet.Maintenance: Not graded. No commit or release date has been recorded.
How to install one
- Compare the shortlist. Code-Graph-RAG leads this topic. Open two or three listings and compare their tools, transport, and required API keys.
- Copy the config for your client. Each listing has an install guide, for example Claude Desktop, Claude Code, Cursor, or VS Code. The client guides show where each config file lives.
- Scope credentials, then restart. Use the narrowest token or folder the server needs, restart the client, and check that the new tools appear.
RAG MCP questions
Do I need a vector database to use a RAG MCP server?
Not always. Some servers bundle a local store such as SQLite with vector extensions or an in-memory index. Larger collections usually move to a dedicated vector database; see the vector database topic for servers that connect to one.
Is a RAG server the same as an AI memory server?
They overlap but serve different data. RAG servers retrieve from documents you load in bulk. Memory servers store facts the agent learns during conversations and recall them later. Many teams run one of each.
Can a RAG MCP server keep my documents private?
Yes, if both the index and the embedding model run locally. Servers that call a hosted embedding API send your text to that provider, so read the server's configuration before indexing anything sensitive.
How do I tell if retrieval is working well?
Ask questions whose answers you know and check whether the returned passages contain them. If they do not, try smaller chunks, a different embedding model, or hybrid keyword and vector search before blaming the language model.