The Agent Skills Directory

[SAFE]: The skill implements a restricted execution environment by limiting allowed tools to the runcomfy CLI, which adheres to the principle of least privilege.
[SAFE]: The skill provides secure secret management guidance, instructing users to use environment variables or local configuration files with restricted permissions (0600) for API tokens.
[INDIRECT_PROMPT_INJECTION]: The skill proactively identifies an indirect prompt injection surface where external audio inputs could potentially influence agent behavior.
Ingestion points: The audio URL parameter used in the audio-inpaint and audio-outpaint endpoints in SKILL.md.
Boundary markers: The skill instructs the agent to only ingest audio URLs that the user has explicitly provided for the music generation task.
Capability inventory: The skill possesses the capability to execute runcomfy run commands via the restricted Bash tool.
Sanitization: The documentation provides remediation guidance for the agent to monitor for divergent output that might indicate poisoned input data.

ace-step