Find the book
Find the title, author, or half-remembered idea hiding somewhere in your shelves.
Goodreads is good at keeping a list. Goodreads MCP lets you ask it questions. Which books changed how you think about work? What do you keep returning to? What should you read next?
No password required. Your Goodreads login stays yours.
Most of us read a book, mark it read, and move on. Later, the useful part is hard to recover. This connects your Goodreads history to Claude or Codex so you can search it, look for patterns, and make decisions from your own experience.
Find the title, author, or half-remembered idea hiding somewhere in your shelves.
Notice what you’ve loved, skipped, and returned to over the years.
Get recommendations that start with your taste—not a generic profile.
The setup is a little technical because computers are. The useful part is simple.
Choose Claude or Codex and run one command. A browser window opens on your device.
setup --client claude
setup --client codex
Goodreads is the only place that sees your password. We’ll find the profile and library details we need. Open Goodreads ↗
The setup saves the connection on your computer. Then ask something you actually want to know:
There is a local option and a hosted Vercel option. Both let your AI read the same library; the difference is where the connection runs.
Your connection details stay on your machine.
command: npx
args: -y, @organized-chaos/goodreads-mcp
environment:
GOODREADS_USER_ID
GOODREADS_RSS_KEY
A small hosted endpoint handles requests. Your Goodreads details are sent with each request and are not saved.
url: https://goodreads-mcp-chi.vercel.app/mcp
headers:
X-Goodreads-User-ID: GOODREADS_USER_ID
X-Goodreads-RSS-Key: GOODREADS_RSS_KEY