The Agent Skills Directory

[SAFE]: The skill defines a rigorous process for code refactoring with a focus on safety, reversibility, and verification through characterization tests.
[COMMAND_EXECUTION]: The skill utilizes the standard Elixir build tool mix to execute tests (mix test). This behavior is expected and necessary for the stated goal of verifying that refactoring steps do not break existing functionality.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes code and documentation files.
Ingestion points: External Elixir source files and project documentation.
Boundary markers: Absent.
Capability inventory: Shell command execution via mix test and file system write access (documented in SKILL.md).
Sanitization: Absent.

refactor-code