The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script scripts/scan_credentials.py using the uv tool. This command-line utility is used to perform automated security scans of proposed operation plans before they are shown to the user.
[CREDENTIALS_SAFE]: The included script scripts/scan_credentials.py contains logic to identify a wide range of sensitive credentials, including GitHub Personal Access Tokens, AWS Access Keys, Slack tokens, and private key blocks. It ensures that these secrets are not accidentally exposed in logs or transcript previews by redacting them if detected.
[SAFE]: The skill establishes a "Human-in-the-loop" protocol, requiring agents to state clear targets, commands, previews, and recovery steps for all mutation operations, which enhances transparency and safety for external system modifications.

mutation-gate