Labsco
MCP SERVER

Hardcover

by AlexSowT

Read-only access to Hardcover.app's book, series and reader data, with prompts that keep queries small.

Games, Entertainment & Fandom
Summary
A reading log an assistant can actually query.

Book recommendations from a model are guesses about what exists; pointed at your Hardcover account they become queries against real titles, series and shelves. The read-only design means the worst outcome of a bad query is a wasted call, not a mangled library.

What it is

A community-maintained server over the Hardcover.app GraphQL API. It is read-only and namespaced, so an assistant can look books, readers and series up without any path to writing back. Not affiliated with Hardcover.app.

What you get
  • Tools grouped under three namespaces: `books.*`, `users.*` and `series.*`
  • Resources such as `hardcover/tag-categories` for the vocabulary behind the data
  • Prompts such as `hardcover/fantasy-this-year` as ready-made starting queries
  • `limit` and `offset` arguments on queries — the API rate-limits, and the docs suggest starting at `limit<=10`
  • Read-only by construction: nothing here changes your Hardcover data
Requirements

Python 3.11+ and a Hardcover API key from hardcover.app/account/api. Export it as `HARDCOVER_API_KEY` including the leading `Bearer ` prefix — the server refuses to start without it, and a missing prefix is the usual cause of a 401. Install with `uv pip install -e .` or pip, then run `uv run src/hardcover_mcp/main.py`. STDIO only.