crewscore
Last released
Find the guardrails your AI agent prompt forgot: which of eight failure modes - prompt injection, hallucination, runaway cost, missing human approval and more - the text never guards against. Offline, deterministic, explainable per-rule findings, with a CI gate.