swarm
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill aggregates reports from multiple parallel workers, creating a surface for indirect prompt injection if a worker's output contains malicious instructions.
- Ingestion points: Worker results and terminal output (Phase C).
- Boundary markers: The process is governed by a 'done predicate' and specific report requirements defined in Phase A.
- Capability inventory: The skill can spawn multiple subagents (general-purpose) and manage project worktrees.
- Sanitization: The instructions explicitly advise against pasting raw worker dumps and require a compact, evidenced result table (Phase C).
Audit Metadata