setup-engine
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted data from external web sources (engine migration guides, changelogs) and local files (design/gdd/game-concept.md) to generate project configuration and documentation. This creates an indirect prompt injection surface where instructions hidden in the data could influence agent behavior.
- Ingestion points: design/gdd/game-concept.md, WebSearch results, and WebFetch content from official documentation URLs.
- Boundary markers: None. The instructions do not specify the use of delimiters when interpolating external content.
- Capability inventory: Write, Edit, Task, WebSearch, WebFetch.
- Sanitization: None. The skill populates CLAUDE.md and technical-preferences.md with extracted data.
- Mitigation: The workflow mandates user review and confirmation (via AskUserQuestion) before any modifications are committed to the file system.
Audit Metadata