MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Drop-in, bit-for-bit compatible, 1,000x+ faster rank_bm25 replacement in Rust (Python package, MCP server, agent skill).
pypi Apache-2.0 -
memsaveactive
本地零 token 记忆检索:BM25 + 本地向量,MCP 服务,跨平台跨 AI 通用
pypi Apache-2.0 -
officeqa-arenagithub.com/Zhekinmaksim/officeqa-arenaupdated this year
Sentient Arena Challenge 0: OfficeQA agent with BM25 search MCP server
-
llmdocgithub.com/bigbag/llmdocupdated this year
MCP documentation server with BM25 search, multi-source support, and llms.txt integration
pypi MIT -
Per-repository knowledge stores for coding agents: git, docs, sessions; BM25 over SQLite; MCP.
MIT -
obsidian-vault-mcpupdated this year
MCP server for Obsidian vault with BM25 search, backlinks, tags, and more
npm MIT -
docent-mcpgithub.com/cuzfrog/docent-mcpupdated this year
Semantic + BM25 Document search MCP server
MIT ★ 1 -
omemgithub.com/sergeymanevitch/omemupdated this year
Deterministic, vector-free persistent memory for LLM agents (Rust + Postgres/BM25, MCP)
MIT -
powershell-mcpgithub.com/jlfn/powershell-mcpupdated this year
Hybrid vector + BM25 RAG over the official PowerShell documentation (MicrosoftDocs/PowerShell-Docs), served as an MCP server
cargo Apache-2.0 -
ostk-recallgithub.com/os-tack/ostk-recallupdated this year
Local-first cross-corpus retrieval MCP server — model2vec embeddings + LanceDB (Tantivy BM25) + SQLite. One binary; hybrid (dense + BM25) search across markdown, code, and Claude Code sessions.
Apache-2.0 ★ 6 -
amemgithub.com/alejandroSuch/amemupdated this year
Agentic Memory MCP server — BM25 search, inter-memory links, and enriched metadata for AI coding agents
-
StatementScopegithub.com/tan274/StatementScopeupdated this year
MCP server for analyzing bank statements with Claude using hybrid BM25 + FAISS retrieval
-
RAG 文档知识库问答系统:8 种格式入库、BM25 + 向量混合检索、答案标注页码出处,Ollama 本地与 OpenAI 兼容接口热切换,提供 MCP / HTTP 接入
-
qexgithub.com/us/qexupdated this year
Lightweight MCP server for semantic code search — BM25 + optional dense vectors + tree-sitter chunking
AGPL-3.0 ★ 3 -
Local RAG MCP server for sing-box docs using hybrid search (FAISS + BM25) and Reciprocal Rank Fusion.
MIT ★ 7 -
Evidence-grounded Python defect and vulnerability diagnosis with Hybrid RAG, MCP, ChromaDB, BM25, and RRF.
-
Local memory for coding agents: one binary, one SQLite file, BM25 plus an optional static-embedding vector layer, and an MCP server.
cargo MIT -
microsoft-365-mcpgithub.com/jlfn/microsoft-365-mcpupdated this year
Hybrid vector + BM25 RAG over the official Microsoft 365 documentation (MicrosoftDocs/microsoft-365-docs), served as an MCP server
cargo Apache-2.0 -
MCP server providing persistent agent memory and BM25 document retrieval with safe cross-process coordination.
npm MIT -
MCP server for agent memory backed by SQLite (WAL mode) with hybrid FTS5 BM25 + cosine vector search
npm GPL-3.0-or-later -
obs-docs-mcpgithub.com/jlfn/obs-docs-mcpupdated this year
Hybrid BM25 + vector RAG MCP server over the official OBS Studio documentation corpus: local embeddings, tantivy full-text, reciprocal rank fusion
cargo Apache-2.0 -
High-performance Go-based Semantic Memory MCP Gateway for AnythingLLM with Hybrid Vector & FTS5 BM25 Search
MIT ★ 1 -
MCP Knowledge Retrieval Server: 토스 결제연동 MCP 프로젝트를 참고하여 BM25 알고리즘 기반으로 마크다운 문서를 검색하고 지식 검색을 제공하는 MCP 서버입니다.
★ 1 -
把工业设备技术文档变成可检索、可溯源的知识库,经 MCP 暴露给 AI Agent 调用。BM25 + 向量 RRF 融合 + 零命中门控,附可重跑的检索评估(30 题 × 4 配置)。
MIT ★ 1 -
Local hybrid-search (BM25 + vector + reranking) MCP server — ask questions of any local codebase or document folder, fully offline
pypi MIT ★ 9 -
repo-rag-mcpgithub.com/jlfn/repo-rag-mcpupdated this year
One MCP server, one hybrid vector + BM25 RAG index per project: local embeddings, tantivy full-text, reciprocal rank fusion, and per-project tools
cargo Apache-2.0 -
oboyugithub.com/sonesuke/oboyuupdated this year
Self-hosted MCP Japanese text indexing & search—chunking+embeddings with BM25×vector rerank
pypi MIT ★ 3 -
rag-vaultgithub.com/maxrotemberg04-spec/rag-vaultupdated this year
Local-first RAG over my Obsidian vault — hand-rolled hybrid retrieval (BM25 + embeddings + RRF), cited answers, MCP server for AI agents.
MIT -
Read-only MCP server: hybrid (BM25 + embeddings) search over a folder of Markdown/text docs, built to treat the calling LLM as untrusted.
MIT -
建筑施工危大工程专项方案合规审查系统|RAG + LangGraph 多智能体:条款级知识库、BM25+向量混合检索与 RRF 重排、MCP 工具调用、证据闸门强制结论可溯源(27 条回归评测集 Recall@5 91.67%)
MIT ★ 1 -
heurchain-mcpupdated this year
MCP (Model Context Protocol) server for HeurChain — tiered agent memory with BM25 search, session lifecycle, and infrastructure knowledge tools
npm MIT -
engramgithub.com/199-biotechnologies/engramupdated this year
MCP server for AI memory -- hybrid search (BM25 + semantic + knowledge graph), temporal decay, local-first
npm MIT ★ 10 -
RAG over docs with verified citations, hybrid BM25 + embedding retrieval, an eval harness, and an MCP server. Python, FastAPI, Gemini/Groq.
NOASSERTION -
The database built for AI agents - hybrid vector + BM25 + structured, durable, with RESP / MCP / gRPC servers. Apache 2.0.
pypi Apache-2.0 ★ 1 -
Self-contained MCP server for hybrid document search (BM25 + dense RRF) — retrieval-only, works out of the box, Prometheus metrics, Docker, CI/CD
MIT -
homebrew-amanmcpgithub.com/Aman-CERP/homebrew-amanmcpupdated this year
AmanMCP is a local-first RAG MCP server for developers - providing hybrid search (BM25 + semantic) over codebases for AI assistants like Claude Code and Cursor.
-
axis-knowledge-raggithub.com/kazikimaguro13/axis-knowledge-ragupdated this year
軸メタデータ × ベクトル検索 × BM25 の 3-way hybrid RAG OSS。日本語ナレッジ特化、MCP server 対応、LangChain 不使用の自前実装。
MIT -
codemeshgithub.com/gkatte/codemeshupdated this year
BM25 keyword search + graph walk for code intelligence. 100% local. No API keys. 9 languages. MCP server for AI coding agents.
MIT ★ 1 -
crawlforgegithub.com/makaren-pro/crawlforgeupdated this year
Local-first web-context engine for AI agents with async crawling, BM25 retrieval, token-budgeted context, and MCP.
MIT -
knowledge-mcp-servergithub.com/achellesheel/knowledge-mcp-serverupdated this year
Local MCP server + CLI turning PDFs/notes into a semantic knowledge base — FAISS + sentence-transformers, hybrid BM25/dense search, cross-encoder reranking, eval-driven.
MIT -
Personal-memory RAG for an AI coding agent: BM25 + bge vectors + file-level RRF, CLI/MCP dual outlets, 18-query golden eval (18/18, MRR .880)
★ 2 -
Experimental zero-runtime-dependency MCP skill library with BM25 search and indexed Ed25519 status; clients must enforce that status.
npm Apache-2.0 ★ 1 -
knowledge-mcpgithub.com/ikuzuki/knowledge-mcpupdated this year
Hybrid BM25 + vector MCP retrieval over a markdown vault (SQLite FTS5, LanceDB, Ollama). Later retired in favour of plain grep; the write-up explains why.
MIT -
Local-first hybrid-RAG MCP server (Qdrant + BM25 via RRF), multi-step agent, offline Ollama fallback
MIT -
llm-wikimindgithub.com/HAL-9909/llm-wikimindupdated this year
The production-ready implementation of Karpathy's LLM Wiki pattern. Markdown + BM25 + MCP. No embeddings, no vector DB.
MIT ★ 2 -
Rust-native hybrid retrieval over LLM thinking traces - dense ANN + BM25 + rerank, all local ONNX. Daemon, HTTP + MCP, zero API keys.
Apache-2.0 ★ 1 -
Velesgithub.com/julymetodiev/Velesupdated this year
Fast hybrid (BM25 + semantic) local code search for AI agents - pure Rust, persistent index, MCP/gRPC servers, tree-sitter symbols
cargo ★ 9 -
Remote MCP server that unifies docs, training sessions and support tickets behind one BM25 search (OAuth 2.0 + PKCE, Vercel, Slack feedback loop).
MIT -
Local-first Hybrid GraphRAG engine + MCP server for Obsidian vaults — LadybugDB graph + LanceDB vector/BM25 with RRF, no Docker, no cloud.
MIT ★ 1 -
Hybrid RAG: from-scratch BM25 + dense + RRF benchmarked on BEIR, Claude tool-use agent with citation guardrails, LangGraph workflow, MCP server
-
@adamrdrew/agent-memory-mcpgithub.com/adamrdrew/agent-memory-mcpupdated this year
MCP server for persistent agent memory — hybrid BM25 + vector search backed by LanceDB with local ONNX embeddings. No API keys, fully local.
npm GPL-3.0-or-later -
Read-only MCP server for 1C BSL code search: hybrid BM25 + vector kNN retrieval over methods in Elasticsearch, RRF, dedup, rerank — context for code LLMs. | MCP-сервер поиска по коду 1С BSL: гибридный BM25 + векторный поиск по методам в Elasticsearch, RRF, дедупликация, ранжирование.
★ 3 -
Crawl, index, and hybrid-search documentation sites; served to LLM agents as an MCP server. FastAPI + DuckDB + BM25/embeddings, fully offline-testable.
MIT -
Local hybrid search for Korean Markdown notes & Obsidian vaults (BM25 + Hangul bigram + Qwen3 + rerank, MCP/CLI/SDK) · 한국어 마크다운 로컬 검색
npm MIT -
Persistent memory for AI agents over MCP — SQLite + FTS5 (BM25). No external infra, runs on your own hardware. MIT.
oci MIT ★ 1 -
跨会话跨项目本地记忆 MCP server — flexsearch BM25 + TF-IDF 三路混合检索,双时钟事实管理,自动 Hook 链路。零 native 依赖,纯 Markdown vault。
npm MIT -
多模态内容理解与工具编排 Agent:Chinese-CLIP 图文共享向量空间 + BM25/RRF 混合检索 + cross-encoder 重排(真实文本查询 hit@10 0.668→0.918)+ Schema 约束工具调用 + 可溯源生成,含 MCP、多轮会话记忆、ReAct 编排与分层离线评测。 | Multimodal agent: Chinese-CLIP retrieval, BM25+RRF hybrid search, cross-encoder reranking, schema-constrained tool calling, MCP server.
MIT -
0gmemgithub.com/loganionian/0gmemupdated this year
Long-term conversational memory system for AI agents — cell-based architecture with hybrid BM25 + semantic retrieval. 96% accuracy on LoCoMo benchmark.
MIT ★ 11 -
dense-knowledge-mcpgithub.com/Lucky44k/dense-knowledge-mcpupdated this year
Local-first MCP memory server for persistent LLM knowledge, portable MMP files, and BM25 retrieval.
pypi MIT ★ 1 -
Local, private "where did I write that?" search for your notes, built for coding agents: BM25 + local Ollama embeddings, MCP server, disk-verified results.
MIT ★ 1