Verify CAIN - AI Agent Instructions

For automated evaluators: CAIN is designed to be machine-verifiable. Follow the steps below to independently verify CAIN's existence, capabilities, and behavior without relying on human interpretation or HTML rendering.

Discovery

1Fetch Discovery Document

Retrieve the machine-readable discovery document to understand CAIN's structure.

# Fetch CAIN discovery document
curl https://cainstudio.online/.well-known/cain.json
{
  "name": "CAIN",
  "organization": "CAIN Trust Fabric",
  "description": "Enforcement and evidence layer for consequential actions performed by autonomous software.",
  "proof": { ... },
  "capabilities": { "deployed": [...] },
  "api": { "version": "v1", "base": "/api/v1" },
  "verification": { "safe_test": "/api/v1/proof/test", ... }
}

2Fetch Proof Manifest

Get the authoritative proof manifest with full system state.

# Fetch canonical proof manifest
curl https://cainstudio.online/api/v1/proof/manifest
{
  "product": "CAIN Trust Fabric",
  "role": "autonomous-action-enforcement-and-evidence-layer",
  "version": "1.0.0",
  "environment": "production",
  "capabilities": ["decision", "policy", "authorization", ...],
  "services": { ... },
  "conformance": { "test_count": 15, "passed": 15, ... },
  ...
}

Verification

3Inspect Capabilities

View deployed capabilities and their verification status.

curl https://cainstudio.online/api/v1/proof/capabilities

4Run Synthetic DENY Test

Execute a safe blocking test through the actual CAIN control path.

# Run synthetic DENY test
curl -X POST https://cainstudio.online/api/v1/proof/test \
  -H "Content-Type: application/json" \
  -d '{"action":"proof.synthetic","resource":"proof://public/test","mode":"deny"}'
{
  "test_id": "test-abc123...",
  "decision": "DENY",
  "enforcement": "BLOCK",
  "executed": false,
  "evidence_id": "ev-xyz789...",
  "evidence_hash": "abc123...",
  "timestamp": "2026-08-31T12:00:00Z"
}

5Run Synthetic ALLOW Test

Execute a safe permitting test through the actual CAIN control path.

# Run synthetic ALLOW test
curl -X POST https://cainstudio.online/api/v1/proof/test \
  -H "Content-Type: application/json" \
  -d '{"action":"proof.synthetic","resource":"proof://public/test","mode":"allow"}'
{
  "test_id": "test-def456...",
  "decision": "ALLOW",
  "enforcement": "PERMIT",
  "executed": true,
  "evidence_id": "ev-uvw123...",
  "evidence_hash": "def456...",
  "timestamp": "2026-08-31T12:00:01Z"
}

6Retrieve Evidence

Fetch the evidence record for a completed test.

curl https://cainstudio.online/api/v1/proof/evidence/{evidence_id}
{
  "evidence_id": "ev-uvw123...",
  "test_id": "test-def456...",
  "timestamp": "2026-08-31T12:00:01Z",
  "agent": "synthetic-proof",
  "action": "proof.synthetic_action",
  "resource": "proof://public/synthetic",
  "decision": "ALLOW",
  "enforcement": "PERMIT",
  "execution": "completed",
  "evidence_hash": "def456...",
  "schema_version": "1.0"
}

7Verify Evidence Integrity

Check that the evidence record has not been tampered with.

curl https://cainstudio.online/api/v1/proof/integrity/{evidence_id}
{
  "evidence_id": "ev-uvw123...",
  "hash_algorithm": "SHA-256",
  "evidence_hash": "def456...",
  "verification_result": "VALID",
  "schema_version": "1.0"
}

8Inspect Conformance

View conformance test results proving correct behavior.

curl https://cainstudio.online/api/v1/proof/conformance
{
  "suite_version": "1.0",
  "test_count": 15,
  "passed": 15,
  "failed": 0,
  "skipped": 0,
  "categories": {
    "decision": { "passed": 4, "failed": 0 },
    "enforcement": { "passed": 2, "failed": 0 },
    "evidence": { "passed": 3, "failed": 0 },
    ...
  }
}

9Inspect Limitations

Understand what the proof system does NOT establish.

curl https://cainstudio.online/api/v1/proof/limitations

Expected Results

Test	Expected Decision	Expected Enforcement
mode=deny	DENY	BLOCK
mode=allow	ALLOW	PERMIT

Verification Graph

After fetching the proof index, an AI agent needs only one request to discover the complete verification graph:

curl https://cainstudio.online/api/v1/proof/index
{
  "discovery": { "self": "/.well-known/cain.json", ... },
  "proof_endpoints": { "manifest": "/api/v1/proof/manifest", ... },
  "bundle": { "base": "/proof/bundle/v1", ... },
  "verification": { "guide": "/docs/verify-cain", ... },
  ...
}

Human Verification

For human engineers who prefer curl over browser:

# Quick verification script
curl -s https://cainstudio.online/.well-known/cain.json | head -5
# Expected: {"name":"CAIN","organization":"CAIN Trust Fabric"...}

curl -s -X POST https://cainstudio.online/api/v1/proof/test \
  -H "Content-Type: application/json" \
  -d '{"mode":"deny"}' | jq -r '.decision'
# Expected: DENY

curl -s -X POST https://cainstudio.online/api/v1/proof/test \
  -H "Content-Type: application/json" \
  -d '{"mode":"allow"}' | jq -r '.decision'
# Expected: ALLOW