auto-build

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill enforces a 'one human approval' gate (Phase 1), ensuring that no automated work begins without explicit user authorization of the task plan. Irreversible or ambiguous steps are explicitly instructed to stop and ask the user for guidance.
  • [SAFE]: Includes a defensive PreToolUse hook (scripts/guard-git-hygiene.sh) that monitors Bash commands. It deterministically blocks high-risk Git operations such as git reset --hard, git clean -f, and bulk staging (git add -A), ensuring the 'clean rollback' guarantee is maintained by machinery rather than just prompt instructions.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to manage Git worktrees, integrate code, and run tests. All operations are scoped to the project directory and follow a structured 'build contract' to maintain repository hygiene.
  • [SAFE]: The implementation uses isolated Git worktrees for parallel tasks, preventing race conditions or accidental contamination of the working branch by multiple concurrent agent processes.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 09:15 AM
Security Audit — agent-trust-hub — auto-build