agent-gen-audio
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends installing a CLI tool named
rawgenaifrom the author's own Homebrew tap repository (WHQ25/tap). This is a vendor-provided resource. - [COMMAND_EXECUTION]: The skill relies on executing the
rawgenaicommand-line tool to generate audio content, which involves passing user-defined prompts and lyrics as command arguments. - [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it takes untrusted user content (music prompts and lyrics) and incorporates it directly into shell commands used for audio generation.
- Ingestion points: User-provided prompts for sound effects and music, as well as lyrics for song generation, are used as inputs in
SKILL.mdand various reference files. - Boundary markers: Commands use double quotes around prompt strings, which provides a basic boundary but does not prevent escape sequences or adversarial instructions from affecting the tool's behavior.
- Capability inventory: The skill uses shell execution for
rawgenai, reads from local files via arguments like--prompt-fileor--lyrics-file, and writes files to the local system using the-ooutput flag. - Sanitization: There are no instructions for the agent to sanitize, escape, or validate the user input before it is passed to the CLI tool.
Audit Metadata