ai-regression-testing
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill serves as a best-practices guide and template repository for implementing regression tests. It emphasizes automated verification to overcome systematic AI blind spots.
- [COMMAND_EXECUTION]: The skill defines a
/bug-checkworkflow that instructs the agent to execute standard development commands, specificallynpm run testandnpm run build. These are expected operations within a development-oriented skill and do not involve unauthorized privilege escalation or persistence. - [INDIRECT_PROMPT_INJECTION]: The testing patterns described create a surface for processing code changes and API responses, which is an inherent part of the development lifecycle.
- Ingestion points: The workflow ingests API response data via
parseResponseand evaluates the project's source code during the build and test phases. - Boundary markers: None explicitly defined in the provided code snippets.
- Capability inventory: The skill utilizes the agent's ability to execute shell commands (
npm run test,npm run build) and write new test files. - Sanitization: Not applicable in this context as the skill is designed for internal project verification rather than processing untrusted external inputs.
Audit Metadata