ai-podcast-creation
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill provides templates for generating podcast scripts from external documents (e.g., in the 'NotebookLM-Style Content' section). This represents an indirect prompt injection surface where instructions hidden within a processed document could influence the agent's behavior.
- Ingestion points: Document content is ingested into prompts for the 'claude-sonnet-45' app in 'SKILL.md'.
- Boundary markers: Absent; untrusted document content is interpolated directly into instructions without delimiters or warnings.
- Capability inventory: The skill utilizes 'infsh app run' for remote model invocation and media processing.
- Sanitization: No input validation or sanitization is demonstrated before processing external content.
- [COMMAND_EXECUTION]: Utilizes the 'infsh' command-line interface for session authentication and running remote audio generation applications. These commands are standard for interacting with the vendor's platform.
- [EXTERNAL_DOWNLOADS]: References installation scripts and image assets hosted on the vendor's official GitHub repository and cloud domain. These references are used for setup and documentation purposes.
Audit Metadata