grimoire-check

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including source code, tickets, specifications, and Architecture Decision Records (ADRs). This creates a surface for indirect prompt injection where malicious instructions could be embedded in the audited data to influence the agent's verdict or status updates.
  • Ingestion points: The skill explicitly reads source artifacts, implementation code, tests, and runtime evidence in SKILL.md.
  • Boundary markers: The skill instructions emphasize deriving a behavior checklist from requirements before looking at implementation summaries to prevent omissions from defining the scope.
  • Capability inventory: The skill is authorized to update status fields in artifacts (tickets, specs, ADRs) via the host's editing tools as described in status-transitions.md.
  • Sanitization: The instructions provide rigorous sanitization guidelines, requiring the agent to 'preserve all non-status content, whitespace, and formatting' and to report every mutation clearly.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 11:09 AM