implement
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The workflow specifies that the agent should run verification commands provided within the task document (docs/task/{ID}-{task-name}.md). This creates a path for executing arbitrary shell commands if the task document is maliciously crafted.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests implementation steps and verification logic from external task documents, which could contain instructions designed to bypass agent constraints or perform unauthorized actions. Ingestion points: Task documents at docs/task/{ID}-{task-name}.md. Boundary markers: None. Capability inventory: File writing (applying implementation steps) and shell command execution (verification commands). Sanitization: No process for validating or escaping content from the task document is described.
Audit Metadata