# canaisee

> A free, public diagnostic for the post-browser web. Enter any URL and
> canaisee grades it on how readable it is to AI agents: a single headline
> grade (A+ through F) backed by a two-tier Hygiene + Frontier breakdown.

canaisee is open-source and built by [Evangent](https://evangent.org). Every
score traces back to a specific, reproducible check; the rubric is published
and versioned, currently v1.3.

## Core resources

- [Homepage](https://canaisee.com/)
- [About the tool](https://canaisee.com/about)
- [Rubric v1.3 (human-readable)](https://canaisee.com/rubric)
- [About our crawler](https://canaisee.com/about-bot)
- [Markdown mirror of this page](https://canaisee.com/index.md)
- [A2A agent card](https://canaisee.com/.well-known/agent-card.json)
- [MCP server card](https://canaisee.com/.well-known/mcp.json)
- [Sitemap](https://canaisee.com/sitemap.xml)
- [Source code on GitHub](https://github.com/8gara8/agentread)

## How the grade is computed

canaisee v1.3 uses a two-tier composite:

- **Hygiene** (85% for content sites, 75% for interactive) — table stakes:
  crawler accessibility, content readability, semantic structure.
- **Frontier** (15% / 25%) — ahead-of-the-curve signals: `Accept: text/markdown`
  negotiation, `/.well-known/mcp.json`, WebMCP tool declarations, rich
  structured data, `Content-Signal` response headers, A2A agent cards, plus
  the older `/llms.txt` and `.md` mirror signals.

A site with perfect hygiene and zero frontier lands at B (85). Frontier
features push into A and A+.

## Policies for AI agents

- Training, search retrieval, and user-initiated fetches are all allowed.
- `robots.txt` names every major AI crawler explicitly and grants access.
- The canonical machine-readable policy lives at
  [/ai.txt](https://canaisee.com/ai.txt).