voiceover

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands to manage Git workspaces (git worktree add), scaffold project files (pnpm fraimz scaffold), and manage GitHub Pull Requests (gh pr create). These commands use placeholders like <flow-id>, which are internally managed by the agent to ensure isolated development environments.
  • [PROMPT_INJECTION]: The skill possess an indirect prompt injection surface as it ingests feature requests to generate executable code and documentation. However, it implements a strong security boundary by requiring explicit human approval of the narration script in Phase 1 before any workspace setup (Phase 2) or automated coding (Phase 3) occurs. This design ensures that the agent's actions remain aligned with user intent.
  • [EXTERNAL_DOWNLOADS]: Standard project synchronization is performed via git fetch origin dev. These operations target the project's own repository and are necessary for the skill's primary function of maintaining a clean worktree.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 02:20 AM
Security Audit — agent-trust-hub — voiceover