Phinq

An open-source runtime governance layer that classifies and controls AI agent tool calls by risk level, enforcing human approval for irreversible actions with tamper-evident audit logging.