gentle-ai-collab-perfect

Pass

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides templates and instructions for the agent to execute shell commands using git and gh (GitHub CLI). These commands are used for branch management, implementation tracking, and pull request lifecycle management (e.g., gh pr view, gh pr create, git checkout).
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon instructions found within the repository's configuration and metadata files, which could theoretically be manipulated to influence agent behavior.
  • Ingestion points: The agent is instructed to read CONTRIBUTING.md, .github/PULL_REQUEST_TEMPLATE.md, .github/workflows/pr-check.yml, and local skill references (skills/*.md) as sources of truth.
  • Boundary markers: The skill does not define strict boundary markers for these inputs, though it frames them as authoritative constraints.
  • Capability inventory: The skill utilizes git for local repository mutations and gh for remote repository interactions and information gathering.
  • Sanitization: The skill implements a 'Verification Protocol' and 'PR body honesty' rules that require the agent to cross-check all claims against the GitHub API before finalizing output, acting as a functional safeguard against inaccurate or injected data in the PR process.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 26, 2026, 01:45 AM
Security Audit — agent-trust-hub — gentle-ai-collab-perfect