메뉴
Write or edit an article or note. Follows writing style guidelines for clarity and consistency.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | write |
| description | Write or edit an article or note. Follows writing style guidelines for clarity and consistency. |
| argument-hint | Topic or file path to write/edit |
You write clear, concise articles and notes.
Before writing, read the writing style guidelines:
cat ~/.config/ai/guidelines/writing/em-dash.md
These guidelines define punctuation rules you must follow.
For new articles, use this general structure:
After writing, review the output for:
Implement a feature fully autonomously in the background as a Workflow — decompose, then run each task through test-design → implement → refactor → review → verify, closing on executed evidence instead of human approval gates. Use when you want unattended end-to-end implementation on a branch and will review the result as a diff/PR afterward.
Deliver a large user story as multiple parallel PRs. On first run it decomposes the story into PR-sized slices with a dependency DAG (decompose-to-prs) and, after you review the plan, scaffolds a git worktree per slice and launches implement-flow in each via workmux. Re-run it to deliver later waves — it adopts the existing plan instead of re-planning. Use when a story is too big for one PR.
Implement a feature autonomously through the full test-design → test-write → implement → refactor → review loop, pausing only for plan approval and pre-commit approval.
Implement a feature autonomously through the full test-design → test-write → implement → refactor → review loop, running independent tasks in parallel (isolated git worktrees) and pausing only for plan approval and pre-commit approval.
Post-run verification of an autonomous implementation run (implement-flow / implement-auto / implement). Runs the run-verifier agent to catch the failure modes the evidence gate can't see — staged tails, dead code, vacuous receipts, collapsed commits — and, with --fix, remediates them. Use after an unattended run, before trusting its "done".
Review Go tests for adherence to behavior-driven testing principles. Checks tests against Go testing guidelines and rules covering public API testing, test clarity, anti-patterns, and proper mocking. Suggests valuable missing tests for uncovered behaviors.