Come in, this is what yOu Can dO
The ethical invitation protocol for AI agents to interact with websites. Simple. Open. Transparent.
AI agents are everywhere — shopping, booking, applying, searching. But there's no standard way for websites to say what agents can do.
Only says "don't crawl" — no way to say "welcome, here's what you CAN do"
Complex, requires APIs — most websites can't implement it
Chaotic, unethical, no consent — bad for everyone
Security nightmare — users don't know what's happening
One JSON file. Clear permissions. Any website can add it in 5 minutes.
// https://yoursite.com/coco.json { "coco_version": "0.1", "site": "yoursite.com", "welcome": true, "permissions": { "read_listings": true, "make_reservation": true, "purchase": false, "submit_forms": false }, "human_readable": "We welcome AI agents to browse and book. No purchases or form submissions." }
That's it. Agents check coco.json before acting. Clear rules. Ethical AI.
Add a coco.json file to your root with your permissions
Before acting, agents fetch and read your coco.json
Agent only does what's explicitly allowed. No guessing.
Transparent. Ethical. User always in control.
Not "stay out" — but "welcome, here's what you can do"
No action without explicit permission
Agents identify themselves honestly
Users see everything their agent does
One JSON file. Anyone can implement it.
Free, open source, belongs to everyone
The future of AI agents is being written now. Help us make it ethical, transparent, and open.