Crawdad

Crawdad Security provides runtime security for AI agents, detecting prompt injection, credential theft, and data exfiltration in real time. It targets developers, IT, and executives in enterprises and regulated industries. The product is delivered as a local-first, self-hosted solution that inspects every agent request on the machine, with a free tier and enterprise options.

Key features

• Audit trail with session context and forensics
• Continuous red team attack simulation
• AI inventory with model and MCP discovery
• Fleet console for self-hosted management
• Multi-layer detection pipeline with ML classifier
• Zero-knowledge, open benchmark