test-remediation-method
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill performs an automated behavior inventory by reading every test file in the project scope, which could allow malicious instructions in test comments or code to influence agent behavior.
- Ingestion points: In references/consolidate.md (Step 2) and references/audit.md (Step 3), the agent is instructed to read all test files and metrics to produce findings and remediation orders.
- Boundary markers: No specific delimiters or safety instructions are defined to separate the data being analyzed from the agent's control logic.
- Capability inventory: The skill has the ability to move files (git mv), commit changes, and perform recovery operations (git reset).
- Sanitization: The instructions do not describe any sanitization or escaping of the test content before processing.
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of external plugins to support specific testing workflows.
- Evidence: references/consolidate.md contains instructions to tell the user to install the mattpocock/skills plugin from the marketplace.
- Context: The installation requires user interaction and authorization.
Audit Metadata