メニュー
Use when writing or editing technical prose — tech specs, design docs, programmer articles, developer docs, RFCs, ADRs, incident writeups. Do NOT use for code comments, commit messages, or casual chat.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Implement work from a ticket, spec, or inline description; auto-detects the next item from the task folders when no argument is given.
Show status overview and next steps for an .aiwork/ spec. Use when user says "next steps", "what's left", "status", "progress", or wants to see what's planned, unplanned, or incomplete for a task.
Scan current directory (recursively) for Claude Code skills, slash commands, agents, hooks, and plugin/MCP configs, then render an interactive HTML graph of their explicit and implicit relationships. Use when the user says "skills map", "skill graph", "visualize skills", "map skills", "how do my skills relate", or wants to audit a skill collection before publishing.
Turn the current conversation context into a spec. Use when user wants to create a spec (or PRD) from the current context.
Break a plan, spec, or the current conversation into vertical-slice tickets, each declaring its blocking edges.
Structure AI work artifacts under .aiwork/{YYYY-MM-DD}_{slug}/ folders using the project's protocol (triage, research, spec, prd, plan, review, notes, docs/). Use when the user asks to plan, make a spec, triage, research a task, write a report, code review, summarize findings, save the work, document decisions, record this, write that down, persist findings, or references any path under .aiwork/. Also use when starting a non-trivial task that needs a plan or spec before implementation, or when finishing one and capturing notes.
| name | technical-writing |
| description | Use when writing or editing technical prose — tech specs, design docs, programmer articles, developer docs, RFCs, ADRs, incident writeups. Do NOT use for code comments, commit messages, or casual chat. |
Open every document and section with the conclusion. Support afterward. No warm-up, no restating the section title.
Buried:
Session cookies currently expire after 24 hours, causing frequent mobile logouts. We evaluated refresh tokens, extended sessions, and device-bound credentials. We recommend refresh tokens with a 30-day rolling window.
Lede first:
We're replacing 24-hour session cookies with 30-day rolling refresh tokens to fix frequent mobile logouts. Rejected alternatives (extended sessions, device-bound credentials) are discussed below.
Who does what is load-bearing. Reach for passive only when the agent is genuinely unknown or the object is the deliberate topic.
Passive:
When a request is received, the payload is validated against the schema. If validation fails, a 400 is returned and the error is logged.
Active:
The handler validates each request against the schema. On failure, it returns a 400 and writes the error to the audit log.
Plain-English first, term in parentheses after. Spell out acronyms on first use.
Jargon-first:
We use CRDTs in the collaborative editor. Conflict-free replicated data types let clients apply changes locally and merge without a coordinator, removing the round-trip latency of the old WebSocket approach.
Accessible-first:
The collaborative editor lets clients apply changes locally and merge without a coordinator, removing the round-trip latency of the old WebSocket approach. The mechanism is a data-structure family called CRDTs (conflict-free replicated data types).
