Sandbox
1 repo for ai-git · Claude Code · SecurityClear

Scan any agent skill — authored or compiled — for prompt injection, secrets, and malicious execution before it touches your agent. Or compile the long tail on the fly.

33