The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run standard Ruby on Rails verification commands such as bundle exec rspec and bundle exec rubocop to validate code changes and ensure quality.
[DYNAMIC_EXECUTION]: The skill involves the generation of new Ruby files, specifically service objects and RSpec test files, which are subsequently executed to verify behavioral parity during the refactoring process. This activity is restricted to standard development templates and follows local project conventions.
[INDIRECT_PROMPT_INJECTION]: The skill processes an existing codebase, which serves as an external data source. It includes several safety measures to mitigate risks associated with untrusted input.
Ingestion points: Project source code files (e.g., controllers, models, and specifications) read by the agent for analysis and refactoring.
Boundary markers: The skill mandates the creation of an explicit 'Stable behavior statement' and a 'Shim decision' to lock the expected behavior before any structural changes are implemented.
Capability inventory: Execution of shell commands for testing (rspec) and linting (rubocop), and the ability to write to the local filesystem within the project directory.
Sanitization: The process enforces a 'One boundary at a time' approach, reversible steps, and continuous verification loops, preventing unauthorized or large-scale unvalidated modifications.

refactor-code