qa-refactoring
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of educational documentation, checklists, and operational patterns designed to guide AI agents and developers in performing safe refactoring.
- [COMMAND_EXECUTION]: The skill includes examples of utility scripts for batch refactoring and feature flag management (e.g.,
run-codemod-batched.sh,rollback-migration.sh). These scripts are transparently documented, scoped to the skill's primary purpose, and correctly use environment variables for sensitive API keys. - [EXTERNAL_DOWNLOADS]: The documentation references standard industry tools and libraries (such as OpenRewrite, Stryker, and SonarQube). All external links and package recommendations point to official documentation, trusted organizations, or well-known open-source services.
- [INDIRECT_PROMPT_INJECTION]: The skill explicitly addresses the risk of LLM agents expanding scope or altering behavior during refactoring tasks. It provides robust mitigation strategies, including the use of immutable characterization tests (acceptance gates) and mutation testing to verify safety nets before changes are made.
Audit Metadata