task-management
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external 'approved plan documents' to generate tasks and commands, creating an attack surface for indirect prompt injection.
- Ingestion points: The skill reads an 'approved plan document' in Phase 1 to extract implementation steps and verification logic.
- Boundary markers: The skill mandates multiple 'STOP' gates requiring user confirmation of the task list (Phase 1), progress reviews every three tasks (Phase 3), and batch reviews (Phase 4).
- Capability inventory: The skill is capable of file system modifications, arbitrary shell command execution via task verification fields, and subagent dispatch via the 'Agent' tool.
- Sanitization: The skill relies on manual user approval and human-in-the-loop checkpoints rather than automated sanitization of the ingested plan content.
- [DYNAMIC_EXECUTION]: The skill executes shell commands defined in the 'Verification' field of the task template. These commands are dynamically extracted from the external plan document and executed at runtime to confirm task success.
Audit Metadata