The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides access to tools like bash, git, and file operations. These are intended for legitimate security functions such as running cryptographic tests, inspecting logs, and managing audit records within a development environment.
[SAFE]: The skill's instructions strictly enforce security best practices, including mandatory cryptographic identity proofs, fail-closed authorization, and tamper-evident logging. No patterns of prompt injection, data exfiltration, or malicious obfuscation were identified.

agentic-trust