scraped-docs-mcp

Streamable HTTP MCP server for searching scraped markdown docs from a prebuilt SQLite FTS5 database.

Build the database

python scripts/build_db.py

The script reads local scraped/*.md files recursively and writes data/db/docs.sqlite. The scraped/ directory is intentionally ignored and not committed.

Run locally

npm ci
npm run build
$env:DOCS_DB_PATH = "$PWD\data\db\docs.sqlite"
npm start

Health check:

curl http://localhost:3000/health

MCP endpoint:

http://localhost:3000/mcp