skill-install-dry-run-planner

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run in scripts/check.py and scripts/plan_dry_run.py to execute internal validation scripts and the wagents CLI tool.
  • Evidence: subprocess.run(command, check=False).returncode in scripts/check.py and subprocess.run(command, cwd=REPO_ROOT, text=True, capture_output=True, check=False) in scripts/plan_dry_run.py.
  • Context: These calls are used to invoke Python-based validation logic and the environment's skill management CLI (wagents) in a restricted --dry-run mode. The commands are constructed from static lists or controlled arguments, minimizing injection risks.
  • [PROMPT_INJECTION]: The evals/evals.json file contains negative test cases specifically designed to ensure the agent refuses to perform live installations even when requested.
  • Evidence: The negative-apply-request and apply-gate-refusal evals verify that the agent declines wagents skills sync --apply or npx skills add commands, reinforcing the dry-run-only scope of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 06:14 AM
Security Audit — agent-trust-hub — skill-install-dry-run-planner