The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local Python script tools/music_gen.py to interface with the ACE-Step 1.5 model. This is standard behavior for AI agent skills managing local tooling.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided natural language through the --prompt and --lyrics flags. This creates a surface where embedded instructions could theoretically influence the music generation process.
Ingestion points: User-supplied prompts and lyrics passed as arguments to the music generation script in SKILL.md.
Boundary markers: The documentation demonstrates using a heredoc with quoted delimiters ('LYRICS') when writing to files, which is a security best practice that prevents unintended shell expansion.
Capability inventory: The skill triggers a local Python execution that communicates with RunPod serverless endpoints.
Sanitization: No explicit input sanitization is mentioned in the documentation, though the shell usage patterns provided are secure.

acestep