- Home
- Skills
- Ratacat
- Claude Skills
- Workflows Work
workflows-work_skill
- Python
24
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill ratacat/claude-skills --skill workflows-work- SKILL.md11.8 KB
Overview
This skill executes a work plan (plan, specification, or todo file) to ship complete features reliably and quickly. It emphasizes clarifying requirements up front, following existing code patterns, continuous testing, and incremental commits to maintain quality. The goal is finished, review-ready pull requests with clear documentation and screenshots for UI work.
How this skill works
The skill reads the input work document, asks clarifying questions if anything is ambiguous, and requests permission to proceed. It sets up a safe working branch or worktree, breaks the plan into prioritized TodoWrite tasks, and iterates through tasks: implementing code, writing and running tests, and updating the plan checkboxes. It enforces incremental commits, linting, quality reviewers when needed, screenshot capture for UI changes, and a final PR with a descriptive body and links.
When to use it
- Implementing a new feature from a spec or todo file
- Completing bug fixes that must follow project patterns
- UI work that requires screenshots and design verification
- Parallel development when you want to avoid committing to the default branch
- Any work where continuous testing and clear task tracking matter
Best practices
- Ask all clarifying questions before coding and get explicit approval to proceed
- Work on a feature branch or worktree; never commit to default branch without permission
- Break the plan into small, testable TodoWrite tasks and mark tasks in the original plan as completed
- Commit only complete, meaningful units with tests passing; avoid 'WIP' commits as a habit
- Run tests and linting frequently and use reviewer agents only for large or risky changes
- For UI work, capture before/after screenshots and include uploaded URLs in the PR body
Example use cases
- Build a new authentication flow from a specification and create tests and UI screenshots
- Fix and refactor an email validation routine while keeping commits small and descriptive
- Implement a dashboard component following Figma, iterating with design-sync and uploading snapshots
- Complete a cross-cutting refactor that requires reviewer agents for performance and security checks
- Take a TODO list for a small feature, break it into tasks, and ship a PR with coverage and conventional commit messages
FAQ
Ask clarifying questions immediately and get user approval to proceed; it’s better to pause and clarify than to build the wrong thing.
When should I run reviewer agents?
Only for large refactors (10+ files), security-sensitive or performance-critical changes, or when the user requests a detailed review. For routine features, tests and linting are usually sufficient.