The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses run_command to execute docker and curl utilities. These are used for legitimate environment setup (managing local database containers) and functional testing (querying local API endpoints).
[SAFE]: Network operations are restricted to communication with local services running on localhost (ports 3000, 3001, and 27017). No external network exfiltration or communication with remote domains is attempted.
[SAFE]: The skill processes data from the applications under test. While this represents a theoretical surface for indirect prompt injection, the risk is negligible as the interaction is limited to a local development environment controlled by the user.
[SAFE]: No hardcoded credentials, sensitive file access, or persistence mechanisms were found. The skill's behavior is entirely consistent with its stated purpose of functional parity verification.

auditing-parity