ai-podcast-creation
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references external resources for installation and configuration, including the
beltCLI and various utility skills from repositories under theinference-shandbelt-shGitHub organizations. - [COMMAND_EXECUTION]: The skill uses the
beltCLI (via a scoped Bash tool) to execute remote applications for audio synthesis, script generation, and media merging. This is the core intended functionality of the skill. - [PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by ingesting untrusted user data (e.g., documents for podcast conversion) and passing it to large language models (Claude) and text-to-speech engines.
- Ingestion points: Document content and custom scripts processed in
SKILL.mdworkflows. - Boundary markers: Not explicitly defined in the example prompt templates.
- Capability inventory: Execution of
beltcommands for inference and media processing. - Sanitization: Content is interpolated directly into model prompts without specific filtering or escaping mechanisms.
Audit Metadata