recipe-build
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes shell commands to verify the presence of project documentation and task files.
- Evidence: Uses
ls -la docs/plans/*.mdandls docs/plans/tasks/*.mdto check prerequisites. - [PROMPT_INJECTION]: Vulnerable to indirect prompt injection through the ingestion of external task and plan files.
- Ingestion points: Reads work plans and atomic tasks from
docs/plans/anddocs/plans/tasks/(SKILL.md). - Boundary markers: Absent. The skill does not use delimiters or instructions to prevent the agent from obeying commands embedded within the processed files.
- Capability inventory: The skill can invoke sub-agents for code execution (
task-executor), perform quality checks, and execute git commits. - Sanitization: Absent. There is no evidence of validation or filtering of the content read from the task files before they are passed to sub-agents.
Audit Metadata