grok-voice
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill manages the Pixazo API key by storing it in a local configuration file (~/.pixazo/api-key). It instructs the agent to use chmod 600 to secure the file, which follows best practices for local credential management.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text prompts and remote audio URLs, which represents a vulnerability surface for indirect instructions.
- Ingestion points: Untrusted data enters the context via user-provided prompts and external audio links referenced in SKILL.md.
- Boundary markers: No delimiters or safety warnings are included in the prompt construction examples to distinguish data from instructions.
- Capability inventory: The skill uses network utilities (curl, requests, fetch) and file system writing, which could be leveraged if an injection is successful.
- Sanitization: The skill lacks explicit sanitization or validation for external content.
- [REMOTE_CODE_EXECUTION]: The documentation suggests an npx command to install additional components from the author's official repository. This involves downloading and executing remote code as part of a standard installation procedure.
Audit Metadata