script-judges
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided video scripts and creator transcripts for evaluation.
- Ingestion points: Processes scripts and calibration material provided in the working copy ("Feeding the judges" section in SKILL.md).
- Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between internal instructions and external data.
- Capability inventory: The skill is primarily pure-text; it describes integration with the ClipForge API (
PATCH /api/project/{id}/scripts) for applying rewrites, but lacks arbitrary command execution or broad network access. - Sanitization: There are no mechanisms described for sanitizing or validating the input data before processing.
Audit Metadata