The Agent Skills Directory

[SAFE]: The skill outlines a disciplined refactoring process with explicit guardrails to prevent unnecessary architectural changes and ensure behavior preservation. It requires user confirmation before implementing changes, which serves as a safety checkpoint.
[COMMAND_EXECUTION]: The workflow involves executing commands for validation, such as tests, typechecks, and linters, within the local environment as specified in Workflow step 7.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process source code and version control information, which are externally influenced inputs.
Ingestion points: SKILL.md Workflow steps 1 and 2 involve reading git status outputs, diffs, and existing source code files.
Boundary markers: The skill does not provide specific instructions for using delimiters to isolate untrusted code content from agent instructions.
Capability inventory: The skill possesses the capability to modify local files (Workflow step 6) and execute shell commands for verification purposes, such as running tests or linters (Workflow step 7).
Sanitization: No explicit sanitization or validation of the ingested code content is performed by the skill instructions.

code-structure-cleanup