# coldread

Track every agent on your site, API, CLI, and MCP.

To add it, give your coding agent this prompt: "Add Coldread to this repo. Read https://coldread.apappas.dev/install.md and follow it."

Not live yet? A cold read: a real AI agent reads your site cold — fetch mode
and browser mode — then grades what it understood: an Agent Readiness Score
out of 100 with a letter grade, per-task grades with verbatim transcripts as
receipts, the fetch-vs-browser mode gap, and one ranked list of fixes with
passive risk findings first.

## Content

- [Home — one agent's visit, the install prompt](https://coldread.apappas.dev/)
- [Product — the dashboard on example data, every surface, a cold read](https://coldread.apappas.dev/product)
- [How it works — getting started, scoring, fixes](https://coldread.apappas.dev/how-it-works)
- [Install — website, CLI and MCP tracking, written for agents](https://coldread.apappas.dev/install.md)
- [Scoring methodology — the full rubric, published openly](https://github.com/andreasjpappas021/coldread/blob/main/agent-readiness-rubric-v1.md)
- [Source](https://github.com/andreasjpappas021/coldread)

## For agents

Reports live at /r/{slug}. The grader identifies itself honestly as
"Coldread/0.1 (Agent Readiness Grader)" and respects robots.txt.
This page is also available as markdown: request / with Accept: text/markdown.