ai-qa-review
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of various industry-standard test runners and mutation testing tools through the CLI. This includes Vitest, Jest, Playwright, pytest, StrykerJS, PIT, cargo-mutants, and mutmut. These commands are part of the standard verification workflow for QA tasks.
- [EXTERNAL_DOWNLOADS]: The instructions reference several external libraries and AI evaluation tools such as promptfoo, deepeval, and ragas. These are well-known tools in the software testing and AI evaluation ecosystems used for establishing quality gates.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external project data, including source code, test files, and local configuration files (e.g.,
.agents/qa-project-context.md). These inputs represent a surface for potential indirect injection if they contain instructions intended to manipulate the agent's review behavior. - Ingestion points: Source code files, test files, and the
.agents/qa-project-context.mdproject-specific configuration. - Boundary markers: The skill provides specific categorical frameworks (Readability, Reliability, etc.) and checklists to scope and structure the agent's analysis.
- Capability inventory: The skill utilizes shell commands for test execution, timing analysis, and mutation score calculation.
- Sanitization: Analysis is performed on static code content using predefined quality dimensions.
Audit Metadata