grill-system-design

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill involves a focused interview process to gather system design decisions from users. While ingesting untrusted user data creates a potential surface for indirect prompt injection, the skill specifically instructs the agent to apply a 'question boundary' from its configuration to mitigate this risk.
  • Ingestion points: User responses during the design interview.
  • Boundary markers: Explicitly references grill-to-spec's question-boundary in step 1.
  • Capability inventory: Orchestrates sub-skills (design-by-contract, grill-to-spec).
  • Sanitization: Relies on the referenced boundary mechanism to filter or delimit user input.
  • [SAFE]: No malicious code, obfuscated URLs, remote downloads from untrusted sources, or persistence mechanisms were identified. The use of relative paths for documentation references is standard for local skill orchestration.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 09:35 AM
Security Audit — agent-trust-hub — grill-system-design