orchestration
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No security issues detected. The skill provides a detailed operational doctrine for AI agent orchestration. It emphasizes cross-vendor verification and structured specs to maintain quality and security when delegating code implementation tasks.\n- [INDIRECT_PROMPT_INJECTION]: The orchestration pattern involves ingesting data (code, diffs, command output) from sub-agents, creating an inherent attack surface for indirect prompt injection.\n
- Ingestion points: Files and tool outputs generated by implementation agents are read by the architect agent in SKILL.md.\n
- Boundary markers: The skill suggests using a fresh context for final reviews, but does not explicitly mandate strict delimiters for external data interpolation.\n
- Capability inventory: The agent is authorized to invoke implementation agents and run verification commands as part of its routing logic.\n
- Sanitization: The doctrine relies on human/agent verification of outputs rather than automated sanitization of external text.
Audit Metadata