parallel
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted task lists and plan items to create subagent briefs, which presents an indirect prompt injection risk.
- Ingestion points: Task descriptions from user input or previous planning phases are used to define subagent goals and scope.
- Boundary markers: While the skill advocates for self-contained briefs, it does not specify technical delimiters or 'ignore' instructions for user-provided task content.
- Capability inventory: The skill has permissions to dispatch subagents, write to documentation files (e.g., 02-DOCS/wiki/sdd/decisions.md), and execute local verification scripts.
- Sanitization: The skill instructions do not include steps for sanitizing or escaping task-specific content before it is passed to subagents.
Audit Metadata