Topic
MCP servers for RAG
Give an agent your own documents to answer from.
Two different things get called a RAG server, and the difference decides what you are installing. One kind indexes your documents: it reads directories, chunks what it finds, and sends the text to an embedding model. The other only queries an index you already built, and never touches your disk at all. Both appear in the same search results.
- Graded servers
- 74
- Grade A
- 41
- Run remotely
- 28
- Reviewed by users
- 0
Before you install one
Three things that separate the servers here, none of which shows up in a description.
- Does it index your files, or only query an index?
- An indexer needs a path and takes everything under it, including the files you forgot were there. A query-only server needs an endpoint and a key. The tool list on each listing page tells you which one you are looking at before you install it.
- Where does the document text go?
- Almost every indexer calls a hosted embedding API, which means the contents of those files leave the machine. That is fine for public docs and a problem for contracts and customer data. If the content is sensitive, look for one that can point at a local embedding model.
- Can it delete as well as search?
- Reindex and drop-collection tools often ship alongside search in the same server. An agent that can drop a collection eventually will, usually while trying to fix something else.
Every listing below has a scan report saying what the server asks for and what it does with it. The rubric is on how we grade.
The RAG servers we would look at first
Reviewed servers rank first, then the ones with real adoption. At most two per publisher, so no account can fill the list. Nothing here is sponsored.
Easy-to-setup local RAG server with minimal configuration
No reviews yetlocal · npm4004.8k/wkFreeCitation-guarded retrieval over 22M Taiwan court judgments and administrative interpretations
No reviews yetgateway-ready320FreeLocal-first agentic RAG with citations - hybrid search, reranking, multimodal document retrieval
No reviews yetlocal · pypi609FreeRepository intelligence, code graph, history, papertrail, and cross-agent memory for coding agents.
No reviews yetlocal · npm20330/wkFreeSearch and explore source-backed SEO, SaaS, and LLM-wiki knowledge from the XKnow knowledge base.
No reviews yetlocal · npm456/wkFreeAcademic paper knowledge-base MCP server: PDF ingest, vector search with reranking, KB management.
No reviews yetlocal · pypi447/wkFreeRead-only offline MCP server over a curated Shopware developer knowledge base
No reviews yetlocal · npm856/wkFreeLocal RAG MCP for markdown documentation. Retrieval only; host synthesizes answers.
No reviews yetlocal · pypi84/wkFreeMulti-provider search broker for AI agents: 14 providers, 12-step extraction, retrieval workflows.
No reviews yetlocal · pypi5FreeQuery a verified document collection: passages that answer a question, with their source.
No reviews yetlocal · pypi2FreeSearch
No reviews yetgateway-readyFreeCheck content rights before RAG, AI input, training or indexing. Signed RSL evidence, not a license.
No reviews yetgateway-readyFree
36 more, graded and ranked
- 13brainatticThe knowledge base your AI reads and writes, under your rules — over MCP, EU-hosted.not reviewedNewC
- 14IWEMarkdown knowledge base as agent memory. Runs against the notes directory it is started in.not reviewedEstablishedA
- 15
DocsMintCloud or self-hosted knowledge for AI agents: hybrid search, reranking, GraphRAG, scoped MCP tools.not reviewedEstablishedA - 16RevoGrid DataGrid MCPToken-free MCP server for structured RevoGrid Core, Pro, and Enterprise knowledge retrieval.not reviewedEstablishedA
- 17Search1APIWeb search, news, page retrieval, sitemaps, and trending topics through Search1API.not reviewedEstablishedA
- 18Google SurfWeb, academic and code search with graph RAG, data lineage, ontology and cross-project schema links.not reviewedEstablishedB
- 19Obsidian Hybrid SearchSearch Obsidian vaults with hybrid full-text, fuzzy, semantic, and graph retrieval.not reviewedEstablishedB
- 20PDFAgentic RAG over one PDF or a whole folder: hybrid search, selective page reads, tables, OCR.not reviewedEstablishedA
- 21
HypermnesiaPersistent memory for AI agents — 36 cited neuroscience mechanisms, local-first, hybrid retrieval.not reviewedEstablishedA - 22Silica CoreRetrieval over the folder the session opened: files, search, read, code_pack, write_note.not reviewedEstablishedA
- 23M3 MemoryLocal-first memory — 100+ tools, 99.2% LongMemEval-S retrieval@10, hybrid search, GDPR, no cloud.not reviewedEstablishedA
- 24lociCross-session memory and hybrid RAG over your notes and docs. Local-first, with cited answers.not reviewedEstablishedA
- 25DataSinkingFull-text Asian financial reports (China, Korea, Japan) as clean Markdown — 6 tools for RAG agents.not reviewedGrowingA
- 26DocPullSync public web sources into cited context packs for AI agents, RAG, and MCP clients.not reviewedGrowingB
- 27Turbo Quant MemoryLocal-first memory and knowledge graph for coding agents. Compact retrieval, no network.not reviewedGrowingB
- 28Memory IndustryMemoryIndustry persistent memory MCP server: BM25+MMR retrieval and audit.not reviewedGrowingB
- 29GutenbergMCP server for Project Gutenberg — 75,000+ public-domain ebooks with full plain-text retrieval.not reviewedGrowingA
- 30SimplepushSend tasks and notifications to people's phones. Answers build a queryable knowledge base.not reviewedGrowingA
- 31MiniragLocal-first RAG MCP server: hybrid search over a folder of your own documentsnot reviewedGrowingA
- 32TibiawikiTibiaWiki knowledge base: attribute queries over creatures, items, NPCs, quests and spellsnot reviewedGrowingA
- 33TinycontextToken-light local memory with SQLite hybrid retrieval for MCP agents.not reviewedGrowingB
- 34Books & Papers MCP ServerFederated search of books and papers, BibTeX/RIS citations, open-access retrieval and reading.not reviewedGrowingA
- 35Machine Library by Space FrontiersMachine Library by Space Frontiers: retrieval across papers, books, patents, and social sources.not reviewedGrowingA
- 36SpomoryExplainable graph-retrieval memory engine with an RL-trained management policy.not reviewedGrowingA
- 37jDataMunch MCPTabular data retrieval. Index CSV/Excel, query rows, aggregate. 99%+ savings vs raw file reads.not reviewedGrowingA
- 38
Zammad MCPEvery Zammad ticket, article, and Knowledge Base operation as one CLI and MCP server - plus anot reviewedGrowingB - 39hubmeshDeterministic multi-hop graph retrieval for RAG. Zero LLM calls in the query path.not reviewedGrowingA
- 40CortexLocal multi-client RAG server over the Model Context Protocol.not reviewedGrowingA
- 41aus-accounting-mcpReviews, six bounded tax worksheets and cited retrieval from your local Markdown folder. Not advice.not reviewedGrowingA
- 42PilPIL: local-first knowledge base for your Instagram saved posts, with a read-only MCP server.not reviewedGrowingA
- 43SEC IntelligenceSEC 10-K/10-Q/8-K search, RAG Q&A, comparison, and anomaly detection with citations.not reviewedGrowingB
- 44CuemapLocal memory and context retrieval for AI agents through CueMap.not reviewedGrowingA
- 45Github RepoMCP server for GitHub repo health, commit summaries, issue triage, and RAG Q&A.not reviewedGrowingB
- 46MemorybankSeeds, governs and navigates a memory_bank knowledge base: gated writes, routing, drift vs codenot reviewedGrowingA
- 47InfobrokerMulti-provider research MCP server: unified web search, fact-checking, personal knowledge base.not reviewedGrowingA
- 48RGX — Snap Router + RyRank + Pricing-TruthTool selection, Cohere-compatible reranking, and on-chain pricing-truth for agentsnot reviewedGrowingB
Or try them without installing any
Add the gateway once and your agent runs search "rag", reads the grades, and calls the one it picks. Nothing to install per server, and an F-graded server cannot be reached through it.
claude mcp add --transport http mcp-market https://gw.mcp.market/mcp
Related
MCP servers for vector databasesMCP servers for PDFs and documentsMCP servers for SQL databasesBest AI and agents MCP serversAll topics