grilling
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data from the environment and user input, creating a vulnerability surface for indirect prompt injection.
- Ingestion points: User-supplied plans, decisions, and ideas; environmental facts retrieved from the filesystem and other tools via sub-agents (referenced in
SKILL.md). - Boundary markers: The instructions lack specific boundary markers or delimiters (e.g., XML tags or triple quotes) to isolate external content from the agent's primary instructions.
- Capability inventory: The agent is authorized to dispatch sub-agents to access the filesystem and environment tools to gather facts (
SKILL.md). - Sanitization: There are no explicit instructions to sanitize, escape, or validate the data retrieved from the environment before incorporating it into the design tree analysis.
- Mitigation: Wrap external content in delimiters with explicit instructions to ignore embedded prompts and sanitize the gathered facts before processing.
Audit Metadata