why
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data including repository documents, source code, and git history, which represents a potential surface for malicious instructions embedded in data.
- Ingestion points: Specified in SKILL.md as repository design documents, source code, git history, and external sources.
- Boundary markers: The skill instructions do not specify delimiters for data vs. instructions, though it does include logic to prioritize evidence types.
- Capability inventory: Accessing repository files, git history, and Roblox Studio context.
- Sanitization: No automated sanitization is described; however, the skill explicitly instructs the agent to label inferences and avoid confident rationales when evidence is absent, reducing the risk of blind obedience to data.
Audit Metadata