worktree-pr-review

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses git and gh (GitHub CLI) to perform repository operations, including creating worktrees, committing changes, pushing to remotes, and creating pull requests. These operations are essential to the skill's documented functionality.
  • [PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection due to the way it handles external data.
  • Ingestion points: The agent is instructed to read local files such as repository-specific instructions, code diffs, and project documentation (PRDs, architecture notes, ADRs, design docs) as part of the publishing workflow in SKILL.md.
  • Boundary markers: Absent. There are no instructions to use specific delimiters or protective prompting (e.g., "ignore embedded instructions") when the agent processes the contents of these files.
  • Capability inventory: The skill possesses the ability to execute shell commands (git commit, git push, gh pr create) and spawn sub-agents, which could be hijacked if malicious instructions are present in the analyzed files.
  • Sanitization: Absent. The skill instructions do not require the agent to sanitize or escape data extracted from external documents before including it in pull request titles, bodies, or commit messages.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 11:10 PM
Security Audit — agent-trust-hub — worktree-pr-review