BlueFlag Expands AI Agent Governance to Unapproved Coding Tools

BlueFlag Security has expanded its developer risk platform to detect and govern unapproved AI coding agents, including those running locally on engineers' laptops outside code repositories. The platform identifies agents through commit and pull request activity, flags AI-approved pull requests lacking human sign-off, and reads endpoint detection data from CrowdStrike and SentinelOne to catch local tools. Two new remediation tools, AccessGuard and DriftGuard, address identity and misconfiguration risks. BlueFlag's own research found unapproved agents in 83% of organizations surveyed, with over two-thirds of agent identities carrying administrator rights.