andrea9293/mcp-documentation-server

npx -y @andrea9293/mcp-documentation-server   # in mcpServers config
npx skills add https://github.com/andrea9293/mcp-documentation-server --skill documentation-server   # agent skill for the REST API

Usage

• •Add documents via `add_document` or drop .txt/.md/.pdf files into the uploads folder and call `process_uploads`.
• •Search across everything with `search_all_documents` (hybrid full-text + vector) or within one document with `search_documents`; fetch broader context with `get_context_window`.
• •Recommended agent path: the REST API at http://127.0.0.1:3080/api/ (e.g. `curl -s -X POST http://127.0.0.1:3080/api/search-all -d '{"query":"your search","limit":5}'`) avoids loading MCP schemas into context.
• •Open the web UI at http://localhost:3080 for dashboard, uploads, and AI search (if GEMINI_API_KEY is set).

Key features

• ✓Runs fully offline — Orama vector DB with local AI embeddings (Transformers.js), no external databases or cloud APIs.
• ✓Built-in web UI / dashboard that starts automatically on port 3080 alongside the MCP server.
• ✓Hybrid search (full-text + vector similarity) with the parent-child chunking pattern.
• ✓Optional AI search via Google Gemini (bring your own key); every MCP tool also exposed as a REST API.
• ✓Published on the MCP Registry and installable via npx; drag & drop .txt/.md/.pdf uploads.

Best for

Offline, local knowledge base and semantic documentation search for AI coding agents (Claude Code, OpenCode, Gemini CLI, Cursor) without vendor lock-in.

Caveats

• ⚠`search_documents_with_ai` requires GEMINI_API_KEY; without it only local embedding-based search tools are available.
• ⚠Changing MCP_EMBEDDING_MODEL requires re-adding all documents (embeddings are incompatible across models).
• ⚠Embedding models are downloaded on first use (~80–420 MB).