grill-me
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided project plans and designs, which constitutes a standard surface for indirect prompt injection. While the instructions are benign, processing untrusted external data carries inherent risks if that data contains embedded instructions.
- Ingestion points: User-provided design specifications and project plans (SKILL.md).
- Boundary markers: The skill does not define specific delimiters to isolate user-provided content from the system instructions.
- Capability inventory: The agent is authorized to "explore the codebase" and "execute the task" based on the resulting understanding (SKILL.md).
- Sanitization: No explicit sanitization or validation of the input plan is implemented in the skill instructions.
Audit Metadata