scenario-refine-loop
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an iterative refinement loop that processes external data from previous tool outputs to guide subsequent actions.
- Ingestion points: Text assets containing verdicts and critiques are retrieved via
asset_download(SKILL.md). - Boundary markers: The instructions do not specify explicit delimiters or boundary markers for the interpolation of verdict text into the prompt-fix logic.
- Capability inventory: The skill utilizes
model_run,asset_analyze, andsearchto perform asset generation and critique (SKILL.md). - Sanitization: No specific sanitization or filtering of the critique text is mentioned before it influences the next generation step.
- [SAFE]: The skill serves as a procedural guide for quality assurance and asset refinement. It recommends the use of official vendor tools and suggests a manual installation process for missing dependencies via a user-facing command (
npx skills add scenario-labs/skills), which maintains user oversight.
Audit Metadata