skills/gildesmarais/dotfiles/slice-pr/Gen Agent Trust Hub

slice-pr

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill facilitates git branch reorganization and PR management using legitimate developer tools like git and the GitHub CLI (gh). No suspicious exfiltration or credential theft patterns were observed.
  • [COMMAND_EXECUTION]: In Step 9, the skill directs the agent to execute the repository's validation scripts and 'repo gate'. This implies the execution of local build or test scripts found within the repository being processed. This is a standard requirement for ensuring code quality in development workflows.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes content from git diffs and commit messages.
  • Ingestion points: Git diffs and commit history are analyzed in SKILL.md (Steps 2 and 8).
  • Boundary markers: None specified for the diff content.
  • Capability inventory: Shell execution of git, gh, and local validation scripts.
  • Sanitization: No explicit instructions for sanitizing commit messages before using them in PR titles or descriptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 01:45 PM
Security Audit — agent-trust-hub — slice-pr