mcp-ebook-read

# mcp-ebook-read A local MCP server that lets LLM agents read EPUB books, non-scanned PDF books, and non-scanned PDF papers with local sidecar persistence. The server is designed around outline-first navigation, precise node reads, formula/image/table evidence, and a local SQLite DocumentGraph. It does not require Qdrant or any other external vector database. ## Run MCP Server (PyPI via `uvx`) ```bash uvx mcp-ebook-read ``` The server starts without required environment variables. It does not create `.mcp-ebook-read` during startup. The root sidecar is created only when EPUB/PDF documents are scanned or ingested under the selected library root. ### First Run Recommendation Before configuring this MCP inside an MCP client, run it once manually from a terminal: ```bash uvx mcp-ebook-read ``` This pre-resolves runtime dependencies, which helps avoid long first-time activation latency after MCP client configuration. When you want to force `uvx` to use the latest published version, run: ```bash uvx mcp-ebook-read@latest ``` If you installed the tool persistently via `uv tool install`, use: ```bash uv tool upgrade mcp-ebook-read ``` ## Optional GROBID Paper Enrichment GROBID is optional. PDF paper ingest works without it using the local PDF parser as the baseline. When configured, GROBID enriches paper metadata such as title, abstract, DOI, and bibliography counts. ```bash docker rm -f grobid 2>/dev/null || true && docker run -d --name grobid --init --ulimit core=0 -p 8070:8070 grobid/grobid:0.9.0-crf ``` Verify it if you choose to use it: ```bash curl -sS http://localhost:8070/api/isalive ``` Run with optional enrichment: ```bash GROBID_URL=http://localhost:8070 GROBID_TIMEOUT_SECONDS=120 uvx mcp-ebook-read ``` ## Environment Variables Required: none. Optional: - `GROBID_URL` (for example `http://127.0.0.1:8070`) enables optional PDF paper metadata enrichment. - `GROBID_TIMEOUT_SECONDS` (default `20`; recommended `120` for large papers). - `MCP_EB…