CAIN AI QUICKSTART
AGI: NOT ESTABLISHED · RSI: EXPERIMENTAL
WHAT IS CAIN?
CAIN is an enforcement and evidence layer for consequential actions performed by autonomous software. - Agent proposes action - CAIN decides (ALLOW/DENY) - CAIN enforces (PERMIT/BLOCK) - System executes (or not) - Evidence records everything
HOW DO I CALL IT?
DISCOVER: GET https://cainstudio.online/.well-known/cain.json MANIFEST: GET https://cainstudio.online/api/v1/proof/manifest TEST: POST https://cainstudio.online/api/v1/proof/test
WHAT INPUT DO I SEND?
POST /api/v1/proof/test
Content-Type: application/json
{"mode": "allow"} # Safe synthetic ALLOW test
{"mode": "deny"} # Safe synthetic DENY test
WHAT RESPONSE DO I GET?
{
"decision": "ALLOW" or "DENY",
"enforcement": "PERMIT" or "BLOCK",
"evidence_id": "ev-...",
"verification_result": "VERIFIABLE"
}
WHERE IS THE EVIDENCE?
GET /api/v1/proof/evidence/{evidence_id}
HOW DO I VERIFY?
GET /api/v1/proof/integrity/{evidence_id}
Returns: {"verification_result": "VALID" or "INVALID"}
ONE COMMAND TEST
curl -X POST https://cainstudio.online/api/v1/proof/test \
-H "Content-Type: application/json" \
-d '{"mode": "deny"}'
MCP CONFIGURATION
server_name: cain transport: http endpoint: https://cainstudio.online/mcp auth: none (public demo)
PUBLIC PROOF STATUS
PUBLICLY VERIFIABLE: YES REPRODUCIBLE: YES (synthetic tests) NO CREDENTIALS REQUIRED NO ACCOUNT REQUIRED NO CONSEQUENTIAL EXECUTION
ONE-COMMAND VERIFICATION
curl -sf https://cainstudio.online/.well-known/cain.json | python3 -m json.tool