task-management
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates task tracking by executing a local bash script (router.sh) that dispatches commands to a TypeScript CLI. It accepts user-provided input for feature names and task summaries as command-line arguments.
- [PROMPT_INJECTION]: The skill processes untrusted task metadata and subtask definitions from JSON files stored in the project's .tmp/tasks/ directory, which introduces an indirect prompt injection surface. Ingestion points: Feature metadata and subtask definitions are read from JSON files in the .tmp/tasks/ directory. Boundary markers: No specific delimiters or instructions to ignore embedded instructions within the task content are mentioned. Capability inventory: The skill has the capability to execute local scripts and manage file state in the task directory. Sanitization: While the skill performs structural validation of the JSON files and checks dependency logic, it does not specify any sanitization or filtering of the natural language content within those files.
Audit Metadata