executing-plans

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because its primary function is to read and execute instructions from a 'plan file'.
  • Ingestion points: The skill reads external data from a plan file as specified in 'Step 1: Load and Review Plan' within SKILL.md.
  • Boundary markers: There are no technical delimiters or explicit instructions provided to the agent to disregard malicious instructions embedded within the plan data.
  • Capability inventory: The agent is instructed to 'Follow each step exactly' and 'execute all tasks', which involves file modifications, code execution, and Git operations (e.g., superpowers:using-git-worktrees).
  • Sanitization: The skill implements mitigation through process logic rather than technical filtering, requiring the agent to 'review critically' and 'STOP executing immediately' if blockers or gaps are identified, which involves human-in-the-loop verification.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 05:31 PM
Security Audit — agent-trust-hub — executing-plans