kickoff
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates as an orchestrator that processes external repository data and user instructions to drive delivery workflows.
- Ingestion points: User intake instructions and repository evidence discovered during the 'Establish the workspace' phase (SKILL.md).
- Boundary markers: The instructions require honoring 'authorization boundaries' and 'repository instructions' (SKILL.md).
- Capability inventory: The skill coordinates implementation via external workers (ship-it), performs version control operations (create-pr), and manages worktrees using 'forest close' (SKILL.md).
- Sanitization: The skill contains an explicit rule to 'Never include secrets or raw sensitive output' in its durable state files (references/episode-state.md).
Audit Metadata