本文へ移動
cccskills
無料GitHub で公開

office

Use whenever a Word document (.docx) in the workspace is to be written, filled in, edited or commented on and the office_edit tool is available: the user watches your edits appear live in the file pane, under your name, and may edit beside you. Covers drafting a new document section by section, revising one, re-wording, tables, lists, charts, pictures and comment threads. Use the docx skill only when the office tools are not available.

インストール方法を見る

含まれるファイル(2)

  • SKILL.md3.7 KB
  • references/word.md3.5 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Writing documents live with the user

The document is open in the user's file pane, or opens there once your first edit to it has run (once per document: a user who closes it is not pulled back). Every office_edit call appears whole, under your name, while the user watches and may type in the same document. Work like a colleague in a shared document, not like a program writing a file.

The workflow

  1. Read first. office_read (outline) before any edit, and again whenever an edit is refused. Target blocks by the id and hash it gives. Quote words with find only for a phrase inside a block, copied from the block view, which gives the block's exact text (the outline shows only its first words). find searches the whole document and nth counts its exact matches from the top, from 1: quote enough words to be unique (the search view ignores case, so its count can differ).
  2. A new document: plant the outline first. A path that does not exist yet: office_edit creates it blank; there is nothing to read first. One office_edit call inserts the title and every section's heading, each followed by a pending block ({"type": "pending", "intent": "…"}) saying what it will hold. The user sees the whole shape at once and can delete a section before you write it.
  3. Then fill one section per call, replacing its pending block ({"pending": "<key>"}). Between calls, say one short line in the chat ("Writing the risks section."). A pending block the user deleted is a section they do not want: leave it out.
  4. Change only what was asked. Re-word with replace on quoted words ({"text": …}), which keeps their formatting; replace a whole block only when it all changes.
  5. When an edit is refused, nothing of the call applied and no one saw any of it. The result names the op and why. When its block changed since you read it, the result shows the block as it stands now: someone edited it, and their edit wins, or an earlier op of the same call changed it. A block that is gone was deleted or moved, by someone or by an earlier op of the call. Read again and work from what is there; never put your old text back over theirs. Ops in one call run in order, each on what the earlier ops left: a block an earlier op replaced, moved or deleted is gone for the later ones (a moved block gets a new id, found in that op's result ids), and one it re-worded has a new hash.
  6. Never touch an open document any other way. No write_file, patch, terminal, python-docx or the docx skill on it: the editor is its only writer, and anything else is refused or lands as a conflicting copy.
  7. Finish with one line in the chat saying what you did and what the user can do next (edit, comment, ask for another pass). The findings belong in the document, not in the chat.

Writing rules

  • Lead with the answer; a reader skimming the first line of each section gets the point.
  • Specifics over adjectives: numbers, names, dates.
  • Short paragraphs; lists for parallel items; a table when the reader compares across rows.
  • Use the document's own styles (headings, lists, table style): never fake a heading with bold text.

See references/word.md for the op formats with examples.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

The Arbor research-coordinator protocol: the OBSERVE -> IDEATE -> SELECT -> DISPATCH -> DECIDE cycle over a durable Idea Tree. Preloaded automatically on /auto-research sessions. The coordinator never edits code or runs commands — executors do that in isolated worktrees; the coordinator steers the tree with idea_tree / dispatch_experiments / merge_experiment.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

The Arbor executor workflow: clone the repo from the git bundle your brief hands you, implement and evaluate exactly ONE hypothesis with the file tools, then report structured results via worker_complete. Preloaded automatically on arbor-executor task workers. Never touch the held-out test split.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

Hard-gated ideation for an Arbor research run. Load at the START of every IDEATE round, before drafting any hypothesis. Enforces the PI mindset (mechanism over knob), the four-question first-principles probe, the kill-filter, and the four-line hypothesis format. Ported from Arbor's idea_drafting + first_principles_probe.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

DECIDE-phase doctrine for an Arbor research run: when to merge, prune, combine, or finalize; how the held-out merge gate works; and the rule that the final report uses TEST scores. Load before deciding what to do with completed experiments.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

Intake for an autonomous research run (Arbor). Use when the user wants to optimize a metric in a repo over many isolated experiments — 'optimize this benchmark', 'improve the model F1 overnight', 'beat the leaderboard'. Discovers the repo/eval/splits, measures the baseline, confirms a Research Contract, then emits a ready-to-send /auto-research command. Does NOT start the run itself.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

You MUST use this before any creative work — creating features, building components, drafting documents, designing workflows, or modifying behavior. Explores your human partner's intent, requirements, and design before any implementation begins.

日本語の概要は準備中です。原文の説明を表示しています。

invergent-ai/surogates282026年10月15日 更新

invergent-ai のスキルをすべて見る

このスキルの問題を報告する