quality-loop
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates an audit of application code to identify features and generate tests, which exposes the agent to instructions embedded within the source code.
- Ingestion points: The agent is instructed to discover features by scanning repository files for routes, workflows, and API endpoints (SKILL.md).
- Boundary markers: The instructions do not define delimiters or provide guidance to ignore embedded instructions found within the processed code files.
- Capability inventory: The agent is authorized to run local unit and end-to-end tests and write a feature inventory spreadsheet to the local disk.
- Sanitization: The skill does not mention any sanitization or filtering of the code content before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill involves running existing repository tests and executing local code verification steps.
- Evidence: Phase 3 and Phase 4 require the agent to run test suites and verify remediation fixes locally.
Audit Metadata