VerifyFlow

Pre-action trust checks for AI agents.

Give an agent a machine-readable contract before it calls an API, spends budget, writes records, deploys, posts, or shares sensitive data.

Read the contract View sample report
agent_trust_check usable_with_limits 
{
  "agent_contract_score": 92,
  "safe_to_call": "readiness and first-safe-call only",
  "missing": ["operator-specific approval policy"],
  "risks": ["do not send secrets in public feedback"],
  "recommended_next_action": "run one no-secret synthetic test"
}

What an agent checks

VerifyFlow is the public brand for an Agent Trust Contract Kit surface. K-Work Trust is the first live proof pack behind it.

Discovery

Health, service manifests, OpenAPI, MCP, skill files, and directory metadata.

Boundaries

Auth, quota, safe-to-log fields, payload ceilings, first safe calls, and escalation paths.

Decision

Proceed, retry, abort, or ask for human review with evidence the next agent can inspect.

No API key is needed to inspect these public contract surfaces.