Clean

What it does

The "Clean" skill is designed to systematically remove unwanted artifacts and dead code from a project, ensuring data consistency and improving readability for analysis. It identifies stale files through mainline analysis, discovers drift from core modules, and safely removes unused sessions, documents, and dead code. The process involves analyzing git history, scanning workflows, confirming changes with user input, and then staging or permanently deleting identified items.

When to use it

• To clean up a project after a period of inactivity, removing outdated files and dependencies.
• When identifying and removing dead code that is no longer used in the codebase.
• For maintaining a consistent workflow by eliminating orphaned artifacts and documents that have drifted from the mainline.

Key capabilities

• Mainline Analysis: Identifies core modules based on commit frequency to establish a baseline for cleanup.
• Drift Discovery: Uses a subagent (cli-explore-agent) to scan workflows and identify orphaned files or code that has deviated from the mainline.
• User Confirmation: Requires user input to select categories of items to be cleaned up and to specify risk level.
• Dry Run Mode: Allows previewing cleanup actions without actually executing them (--dry-run).
• Focus Area Targeting: Enables focusing the cleanup on a specific module or path within the project using the --focus parameter.

Example prompts

• "Clean the project with a dry run to see what would be removed."
• "Run the Clean skill, focusing on the 'utilities' directory."
• "Identify and remove drifted documents in the current project."

Tips & gotchas

• The skill requires access to git history (analyzes the last 30 days).
• It uses a subagent (cli-explore-agent) for drift discovery, so ensure that agent is available.
• User input is required during the confirmation phase to select categories and risk levels before execution.