multi-agent-team
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines an orchestration model focused on clear role separation, file ownership, and controlled communication. The instructions are transparent and align with standard development workflows, with no evidence of obfuscation, remote code execution, or unauthorized data access.
- [INDIRECT_PROMPT_INJECTION]: The skill's orchestration workflow involves passing context and handoff documents between agents, which represents an attack surface for indirect instructions.
- Ingestion points: External task context and cross-stage handoff documents are injected into worker templates in SKILL.md.
- Boundary markers: None explicitly defined in the provided markdown templates to delimit external data.
- Capability inventory: Agents are authorized to edit files and run verification tests within their assigned worktrees.
- Sanitization: No specific content sanitization or validation logic is defined for the transferred context.
Audit Metadata