plan-parallel

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface
  • Ingestion points: The skill ingests untrusted data via the $ARGUMENTS variable in the task block and by reading codebase files such as backend-patterns-reference.md, frontend-patterns-reference.md, and other files identified via the /scout command.
  • Boundary markers: The skill uses XML-like <task> tags to encapsulate user arguments, providing some structural separation.
  • Capability inventory: The skill uses tools like TaskCreate and AskUserQuestion, and invokes other skills via slash commands (/scout, /plan-review, /plan-validate). It explicitly restricts itself to planning and forbids code implementation.
  • Sanitization: No explicit sanitization or escaping of external content is described, but the workflow includes mandatory manual review by the user and automated validation via /plan-review.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:25 AM
Security Audit — agent-trust-hub — plan-parallel