subagent-task-execution
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a procedural framework for managing tasks and does not include any executable code, remote dependencies, or sensitive operations.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external task plans and descriptions, creating an ingestion surface for untrusted data. It mitigates this risk through a mandatory multi-stage review process where specialized subagents verify compliance and quality before task completion. Ingestion points: Task breakdown documents and task descriptions. Boundary markers: Isolation of tasks and mandatory review stages. Capability inventory:
TodoWritetool and subagent dispatching. Sanitization: No explicit content filtering mentioned.
Audit Metadata