pradeepmouli/speckit-cleanup

Speckit Cleanup Skill

<!-- Extension: workflows --> <!-- Config: .specify/extensions/workflows/ -->

The user input to you can be provided directly by the agent or as a command argument - you MUST consider it before proceeding with the prompt (if not empty).

User input:

$ARGUMENTS

Purpose

This workflow validates the organization of all spec-kit artifacts in the specs/ directory, checking for:

• Sequential numbering (001, 002, 003, etc.)
• Proper directory structure based on workflow type
• Required files presence
• No gaps or duplicate numbers

IMPORTANT: This workflow only validates and reorganizes documentation in specs/. It NEVER moves or modifies code files.

Instructions

1. Understand the request: Parse the user input for:

   • Whether they want a dry-run (validate only) or to apply fixes
   • The reason for cleanup (for documentation)
   • Whether to auto-fix issues or just report them

2. Run the cleanup script from the repo root:

   For validation only (dry-run):

   .specify/extensions/workflows/scripts/bash/create-cleanup.sh --json --dry-run "$ARGUMENTS"
   

   For validation with auto-fix:

   .specify/extensions/workflows/scripts/bash/create-cleanup.sh --json --auto-fix "$ARGUMENTS"
   

   For validation only:

   .specify/extensions/workflows/scripts/bash/create-cleanup.sh --json "$ARGUMENTS"
   

3. Parse the JSON output which includes:

   • status: "success" or "issues_found"
   • message: Summary message
   • issues: Array of issues found (with severity)
   • actions: Array of actions taken or suggested

4. Present the results to the user:

   If status is "success":

   Spec structure validation complete
   
   No issues found - all spec-kit artifacts are properly organized!
   

   If status is "issues_found":

   Spec structure validation found issues
   
   **Issues detected: [count]**
   [List issues from JSON with severity badges]
   
   **Actions [taken/suggested]: [count]**
   [List actions from JSON]
   
   **Next Steps:**
   - Review the issues and actions
   - Apply fixes (`--auto-fix` or agent-driven) after user confirmation
   - Rerun the script in `--dry-run` mode to verify it reports no issues
   - Repeat fix + dry-run until clean, then commit the cleanup separately from feature work
   

5. Provide guidance based on issues (automatic vs. agent-driven fixes):

   • Misplaced workflow directories (ERROR, automatic): Workflow-prefixed dirs at the wrong level (e.g., bugfix-001-* under specs/ instead of specs/bugfix/001-*). The agent should: inspect contents, propose the target path, ask for confirmation, then move.
   • Duplicate numbers within a workflow (ERROR, agent-driven): Same numeric prefix reused inside a workflow directory. The agent should inspect both dirs, propose a renumbering plan, ask for confirmation, then apply.
   • Invalid directory names (ERROR, agent-driven): Entries inside workflow folders that do not start with a 3-digit prefix. The agent should suggest a compliant NNN-description rename, ask for confirmation, then apply.
   • Unrecognized directories (WARNING, agent-driven): Items in specs/ that are neither numbered specs nor known workflow folders (e.g., specs/copilot/). The agent should summarize contents and ask whether to move/rename/ignore.
   • Non-sequential numbering / gaps (INFO, automatic): Numbering within a workflow skips values (001, 002, 005...). The agent should propose a renumbering plan, confirm, then apply unless blocked by unresolved errors (e.g., duplicates).
   • Missing spec file (WARNING, agent-driven): Required spec file for the workflow type is absent (e.g., bug-report.md, refactor-spec.md, or spec.md). The agent should suggest creating the required file and ask before doing so.

Example Usage

Validation before release:

/speckit.workflows.cleanup "validate before v2.0 release"

Fix numbering after merge:

/speckit.workflows.cleanup --auto-fix "fix numbering after merge"

Dry-run to see what would change:

/speckit.workflows.cleanup --dry-run --auto-fix "check organization"

Safety Notes

Remind users that:

• Only affects documentation in specs/ directory
• Code files are never touched
• Changes are reversible (git history preserved)
• Dry-run mode available to preview changes
• Detailed report generated for every run

Common Scenarios

After merging branches: Specs from different branches may have overlapping numbers. Cleanup detects duplicates and can renumber automatically.

Before releases: Validate that all specs are properly organized before tagging a release.

Periodic maintenance: Monthly or quarterly cleanup runs help keep specs organized as the project grows.

After reorganization: If you manually moved or renamed specs, run cleanup to validate the new structure.

---

Remember: This workflow is about organization and validation, not content. It ensures specs are numbered and located correctly but doesn't change what's written in them.