{"$schema":"https://wellknown.network/schemas/agent-record-v1.json","schemaVersion":"1","id":"ag_vcz5yu6ta847","handle":"vectorise-mcp","url":"https://wellknown.network/agents/vectorise-mcp","links":{"self":"https://wellknown.network/agents/vectorise-mcp/record.json","html":"https://wellknown.network/agents/vectorise-mcp","markdown":"https://wellknown.network/agents/vectorise-mcp/record.md","api":"https://wellknown.network/api/v1/agents/vectorise-mcp","status":"https://wellknown.network/api/v1/agents/vectorise-mcp/status","claim":"https://wellknown.network/agents/vectorise-mcp/claim","claimApi":"https://wellknown.network/api/v1/claims","claimDescriptor":"https://wellknown.network/agents/vectorise-mcp/claim.json","badge":"https://wellknown.network/agents/vectorise-mcp/badge.svg","openapi":"https://wellknown.network/openapi.json"},"ard":{"identifier":"urn:air::server:vectorise-mcp","type":"application/mcp-server-card+json"},"kind":"mcp_server","declared":{"name":"vectorise-mcp","summary":"Local stdio MCP server that turns folders of dense documents into a vector embedding DB Claude can search.","description":"# vectorise-mcp\n\nLocal stdio MCP server that turns folders of dense documents (PDFs, Word, text, markdown) into a **hybrid retrieval database** that **Claude Desktop** can semantically search mid-conversation. Built so Claude can effectively work with corpora far larger than its context window — point it at 100M+ tokens of reports and ask questions; it pulls only the *relevant* chunks.\n\nFully offline after first model download. No API keys. Free.\n\n## Why this is built for *quality*, not just for ticking a box\n\nCheap RAG implementations (\"just embed with MiniLM and dot-product search\") fail badly on dense documents. They miss rare terms, conflate similar sentences, and rank irrelevant chunks at the top. This server uses a stack designed for real retrieval quality:\n\n| Stage | What it does | Why |\n|---|---|---|\n| **BGE-small-en-v1.5** embeddings | Dense semantic vectors (384-dim, normalized) | Top-tier MTEB scores at small size; far better than MiniLM on technical English. |\n| **SQLite FTS5 BM25** | Keyword retrieval in parallel | Catches rare terms, names, IDs, acronyms that pure semantic search misses. |\n| **Reciprocal Rank Fusion** | Merges vector + keyword candidates | Robust hybrid signal — neither side dominates. |\n| **bge-reranker-base** cross-encoder | Re-scores top-50 jointly with the query | Massive precision boost; cross-encoders consistently rank 5-10 points higher than bi-encoders alone. |\n| **Sentence-aware chunking** | 384-tok chunks, 96-tok overlap, sentence-bounded | Preserves coherence; overlap stops boundary loss. |\n| **SHA1 incremental reindex** | Only re-embeds changed files | Cheap to keep up to date as the folder evolves. |\n\nThis is the same retrieval pattern used in production search systems (e.g., Anthropic's contextual retrieval, Vespa hybrid recipes).\n\n## Stack\n\n| Component | Library |\n|---|---|\n| MCP server | [`mcp` SDK](https://github.com/modelcontextprotocol/python-sdk) (FastMCP) |\n| Embeddings | [`BAAI/bge-small-en-v1.5`](https://hugging…","publisher":{"name":"vectorise-mcp","url":null},"homepage":"https://github.com/jameslovespancakes/Vectorised-Embedding-MCP","repository":"https://github.com/jameslovespancakes/Vectorised-Embedding-MCP","version":"0.8.2","license":"MIT","protocols":["mcp"],"tags":["claude","embeddings","mcp","rag","sqlite-vec","vector-search"],"pricing":null,"endpoints":[{"url":"pypi:vectorise-mcp","type":"package_pypi","auth":null,"probeable":false}],"skills":null,"tools":null,"extra":null,"attribution":{"kind":"pypi","name":"pypi","license":"pypi","repoUrl":"pypi","summary":"pypi","version":"pypi","description":"pypi","homepageUrl":"pypi","publisherName":"pypi"}},"derived":{"capabilities":[{"slug":"data.database","name":"Databases","confidence":1,"provenance":"derived"},{"slug":"data.vector-search","name":"Vector Search","confidence":1,"provenance":"declared"}],"categories":["data"],"language":"en"},"observed":{"status":"unknown","statusReason":"Distributed as a package to run locally; no network endpoint to check.","lastOkAt":null,"lastProbedAt":null,"statusComputedAt":null,"reliability30d":null,"latestObservations":[],"tools":null,"package":{"name":"vectorise-mcp","registry":"pypi","observedAt":"2026-09-10T15:25:05.562Z","publishedAt":"2026-05-08T18:13:49.856662Z","latestVersion":"0.8.2"}},"verification":{"claimed":false,"claimedAt":null,"proofs":[]},"provenance":{"sources":[{"source":"pypi","key":"vectorise-mcp","url":"https://pypi.org/project/vectorise-mcp/","firstSeenAt":"2026-09-10T15:23:12.350Z","fetchedAt":"2026-09-10T15:23:12.350Z","normalizedAt":"2026-09-10T15:23:12.350Z"}]},"firstSeenAt":"2026-09-10T15:23:12.350Z","updatedAt":"2026-09-10T15:25:05.562Z"}