st-generate-tasks
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several bundled CommonJS scripts (scripts/find-strikethroo-root.cjs, scripts/validate-plan-blueprint.cjs, scripts/get-next-task-id.cjs, and scripts/route-task-execution.cjs) to manage the workspace, validate plan files, and handle task metadata. These scripts use standard Node.js filesystem modules to read and write files within the project root.
- [EXTERNAL_DOWNLOADS]: Instructions include a recommendation for the user to run npx strikethroo init if a valid workspace root is not detected. This facilitates project initialization by downloading the official Strikethroo CLI from the NPM registry.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes project plan files, which are external data sources that could contain malicious instructions.
- Ingestion points: Plan body files and workspace configurations (config/STRIKETHROO.md).
- Boundary markers: Absent for plan content; however, the skill relies on strict operating procedures and reference files (references/test-philosophy.md, references/complexity-rubric.md) to guide behavior.
- Capability inventory: File reading/writing within the project directory and execution of local scripts.
- Sanitization: Prescriptive decomposition rules, task minimization targets (20–30% reduction), and validation via helper scripts are used to control output quality and safety.
- [SAFE]: Analysis of the provided scripts and instructions revealed no evidence of data exfiltration, credential theft, or unauthorized remote code execution. The skill behavior is consistent with its described utility for task management.
Audit Metadata