build-inline
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses version control commands to maintain project state and verify progress.
- Evidence: Specifically invokes 'git rev-parse HEAD' and 'git log' throughout the per-task loop and setup phase.
- [INDIRECT_PROMPT_INJECTION]: The skill processes task descriptions and constraints from external project files, which serves as an entry point for instructions.
- Ingestion points: Reads from 'tasks.md', 'docs/agents/project.md', and other project documentation files.
- Boundary markers: Creates task capsules ('task-N-brief.md') and records canonical constraint hashes to scope implementation effort.
- Capability inventory: Capable of modifying project files and executing tests via the integrated 'test-first' sub-skill workflow.
- Sanitization: Employs 'Iron Laws' and a 'Clarify first' step that mandates human intervention whenever a plan gap or ambiguity is detected.
Audit Metadata