coordinator-mode
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The multi-agent orchestration pattern creates a surface for indirect prompt injection through the ingestion of user requests.
- Ingestion points: User requests are processed during the 'Decompose' phase defined in SKILL.md.
- Boundary markers: No technical delimiters are specified for user-supplied content, though the lifecycle mandates a 'Synthesis' phase.
- Capability inventory: Sub-agents dispatched by the coordinator have access to high-privilege tools including 'Bash', 'Write', 'Edit', and 'Agent'.
- Sanitization: The 'Golden Rule: Never Delegate Understanding' serves as a procedural control, instructing the agent to translate user intent into specific directives instead of passing through un-sanitized instructions.
- [SAFE]: The skill does not contain any evidence of hardcoded credentials, malicious obfuscation, or unauthorized external network requests. All referenced tools and workflows are consistent with the stated purpose of agent orchestration.
Audit Metadata