| |
|
|
|
|
|
|
| |
I converted 16 user-invoked skills to commands. Skills cost ~150-300 tokens each per session for descriptions the model uses to auto-route. Commands cost nothing until you type the slash. These 16 are workflows I always trigger deliberately. The auto-routing wasn't earning its keep. This reclaims ~4-5k tokens per session.
Nine skills stayed where auto-routing genuinely helps: debug, root-cause-trace, five-whys, add-tests, frontend-design, humanizer, playwright-js, playwright-py, and pairwise-tests. Pairwise-tests stays a skill because its helper files don't fit a single-file command shape.
For arch-decide, I preserved the upstream MIT LICENSE alongside the command at .claude/commands/arch-decide.LICENSE so attribution stays intact.
|
|
|
A workflow for picking up a specific task (Linear ticket, GitHub issue, todo.org heading, or a described scope) and carrying it through to handoff. Seven phases, three user-approval gates.
The phases in order:
- Claim: mark in-progress, assign, label, verify project.
- Justify (gate 1): benefits, costs, impact, urgency, effort, alternatives, ticket quality.
- Approach (gate 2): root cause, risk, test strategy, migration, flag, commit decomposition, branch name.
- Implement (TDD): red, green, edge cases, and a refactor pass with a nine-point language-agnostic checklist.
- Verify: end-to-end or scripted manual test in the local environment. Playwright for web projects, step-by-step walkthrough for everything else. Never production.
- Ready to commit (gate 3): report, deviations from the approach, any follow-ups spotted.
- Hand off to the Review-and-Publish flow in commits.md.
Cross-referenced from testing.md, commits.md, subagents.md, review-code, brainstorm, arch-design, arch-decide, debug, pairwise-tests, playwright-js, and playwright-py.
|