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.
-
Local code search combining BM25, vector similarity, and cross-encoder reranking. Parses 60+ languages with tree-sitter, runs entirely offline, and returns structured results with file paths, line ranges, and symbol metadata. Built in Rust.
MIT ★ 149 -
Local RAG server for Claude Code, Cursor and other MCP clients: hybrid BM25 + vector search with cross-encoder reranking on by default, 13 MCP tools, 33 file types including code, re-indexing on file changes. One Python process with embedded ChromaDB and FastEmbed ONNX: no Docker, database server or API key; offline after the first model download.
npm MIT ★ 306 -
win-shell-mcpactive
AI-native cross-platform command abstraction layer as MCP Server. Deterministic tools replacing raw shell commands, unified minimal JSON output, handles Windows path/encoding/quoting differences.
npm MIT -
internet-context-mcpgithub.com/vivekvar-dl/internet-context-mcpupdated this year
Read-only MCP server giving AI agents the web as compact, ranked, verified evidence. Local cross-encoder reranker, NLI claim verification, semantic cross-source agreement and contradiction detection. No API keys.
MIT ★ 1 -
internet-context-mcpgithub.com/vivekvar-dl/internet-context-mcpupdated this year
Read-only MCP server giving AI agents the web as compact, ranked, verified evidence. Local cross-encoder reranker, NLI claim verification, semantic cross-source agreement and contradiction detection. No API keys.
npm MIT -
Multi-engine web search + fetch MCP server (DDG, Brave, Tavily, Google AI Mode) with consensus-merge dedup, local cross-encoder reranking, and Groq synthesis
MIT ★ 1 -
@cablate/memory-lancedb-mcpgithub.com/cablate/memory-lancedb-mcpupdated this year
MCP server for LanceDB-backed long-term memory with hybrid retrieval (Vector + BM25), cross-encoder rerank, multi-scope isolation, and memory lifecycle management
npm MIT -
MCP server for LanceDB-backed long-term memory with hybrid retrieval (Vector + BM25), cross-encoder rerank, multi-scope isolation, and memory lifecycle management
MIT -
RAG proof of concept: hybrid retrieval (pgvector HNSW + lexical RRF), cross-encoder reranking, dual embeddings (OpenAI + local GPU/CPU), and an MCP server — Python, PostgreSQL, Docker (NVIDIA/ROCm)
-
pipecat-docs-mcpgithub.com/DTufail/pipecat-docs-mcpupdated this year
MCP server for Pipecat voice AI framework documentation. Search 6,936 chunks from 317 pages via Claude Desktop or Cursor. Hybrid BM25 + FAISS retrieval with cross-encoder reranking. 78.8% MRR, 93% Recall@5, 36ms avg latency.
-
A retrieval engine that refuses to guess. Register git repos and URLs, get grounded, cited answers over MCP and REST. Hybrid search, cross-encoder rerank, Postgres + pgvector.
MIT -
UltraMemorygithub.com/win4r/UltraMemoryupdated this year
Universal AI Agent Long-Term Memory Engine — Hybrid Vector+BM25, cross-encoder reranking, Weibull decay, LLM smart extraction, semantic dedup, tiered content. Zero-ops LanceDB backend.
★ 7 -
sivrugithub.com/sivru/sivruupdated this year
Code search and session observability for coding agents. Hybrid BM25+semantic, optional cross-encoder rerank, self-bench on your own data. npm install -g @sivru/cli
MIT ★ 4 -
Search for 1C (BSL code + configuration metadata) exposed to LLM agents via MCP: local e5/Chroma retrieval, global BM25 + RRF hybrid, optional cross-encoder, file-level eval (nDCG@10 / MRR).
MIT -
The most advanced MCP memory server. Hybrid search (BM25 + jina v5 embeddings + cross-encoder rerank), 100+ languages, temporal knowledge graph, finds any past Cursor or Claude conversation from a vague description, signed automatic updates. Pure Rust, single binary, 100% local, zero API calls.
NOASSERTION ★ 18 -
MCP server exposing hybrid RAG (BM25 + dense + RRF + reranking) over SEC filings — with FastAPI backend, Next.js frontend, and 22-query evaluation.
MIT ★ 2 -
MCP tool server for AI coding agents -- encoding-aware file tools, binary analysis, DAP debugger, SSH/SFTP, process memory, and more
MIT ★ 26 -
context-routergithub.com/mohankrishnaalavala/context-routerupdated this year
Memory-aware context engine for AI coding agents — up to 91% fewer tokens, 17/18 rank-1 across 6 OSS projects. MCP-native, multi-repo, with persistent observations & decisions.
NOASSERTION ★ 11 -
Multilingual decision engine for intents you describe in plain words, no training examples. It picks the matching intent, with a margin that flags doubtful answers; R2Intent v1 also answers yes or no (beta), with probabilities. About 26 ms per message on a GPU (compiled). Python, HTTP or MCP, self-hosted. Library Apache-2.0; models CC BY-NC 4.0.
Apache-2.0 ★ 1