reconcile-review

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub Pull Request comments, descriptions, and diffs. This external content could contain malicious instructions designed to manipulate the agent's behavior during the review process.
  • Ingestion points: The skill uses gh pr view, gh api, and gh pr diff to retrieve content from external PRs into the agent's context as described in SKILL.md.
  • Boundary markers: The instructions include a dedicated 'Security' section that explicitly directs the agent to treat PR content as data, not instructions, and to ignore any text attempting to override the skill's workflow.
  • Capability inventory: The skill possesses significant capabilities, including local file modification, git operations (checkout, commit, push), and posting to the GitHub API.
  • Sanitization: The skill employs a human-in-the-loop mitigation, requiring the agent to present a resolution table and obtain explicit user approval for each specific change before entering the 'Apply' phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:12 PM
Security Audit — agent-trust-hub — reconcile-review