Repository navigation
MCP Knowledge v0.1.0 is the first Docker-first release of a private, explainable local RAG system for collections of roughly 100–1,000 documents.
Highlights
- A polished browser dashboard for managing documents, jobs, access, and backups.
- A search playground that exposes vector rank, lexical rank, RRF score, timing, and source provenance.
- Local embeddings with no hosted AI dependency required to ingest or search.
- Hybrid vector + lexical retrieval over REST and Model Context Protocol (MCP).
- PDF, Word, PowerPoint, Excel, OpenDocument, EPUB, RTF, HTML, XML, JSON, CSV, Markdown, and plain-text ingestion.
- Passphrase-gated dashboard sessions and scoped API keys.
- Single-container persistence, plus documented offline backup and restore.
- Native multi-platform images for
linux/amd64andlinux/arm64.
Run it
docker run --name mcp-knowledge \
-p 127.0.0.1:3000:3000 \
-v mcp-knowledge-data:/app/data \
ghcr.io/raysca/mcp-knowledge:0.1Then open http://127.0.0.1:3000. Use the versioned 0.1.0 tag instead of the moving 0.1 release-line tag to select this release explicitly.
See the README for setup and integrations, the performance guide for reference measurements, and the security guide before exposing the service beyond loopback.
Known limitations
This release is intentionally single-node and single-tenant. It does not include PostgreSQL/S3 backends, Kubernetes deployment, hosted OCR or embeddings, reranking, built-in TLS termination, or general API rate limiting. See the changelog for the complete list.