π° ClawMoat v0.1.0 β Initial Release
Security moat for AI agents
Runtime protection against prompt injection, tool misuse, and data exfiltration.
Features
- Prompt injection detection β pattern matching + heuristic scoring
- Jailbreak detection β DAN, sudo mode, dual persona, encoding bypasses
- Secret scanning β AWS, GitHub, OpenAI, Anthropic, Stripe, SSH keys, JWTs, connection strings
- Policy engine β validates tool calls against configurable security policies
- CLI β
scan,audit,watch,testcommands - OpenClaw middleware β real-time session monitoring
- OpenClaw skill β drop-in skill package
Quick start
npm install -g clawmoat
clawmoat test
clawmoat auditStats
- 20/20 built-in tests passing
- Zero dependencies
- Pure Node.js (>=18)
- MIT licensed
Site: https://clawmoat.com
OWASP Coverage: ASI01-ASI06 of the Top 10 for Agentic Applications (2026)