Formal doc structure
My personal collection of portable SKILL.md agent skills and tooling for Claude Code, Cursor & Codex — currently focused on Traditional Chinese (Taiwan) writing, de-AI editing, and business docs.
npx -y skills add leoluyi/skills --skill formal-doc-structureAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Draft, revise, or restructure formal internal business documents in Traditional Chinese (Taiwan corporate / financial-institution usage) — 簽呈, 會議紀錄, 評估報告, 專案規劃, 採購與廠商溝通, 對主管簡報, 驗收與交付, 跨單位協調. Picks a reader-driven structure per document type and produces a usable draft, not just advice. Trigger when the user asks to write or fix an internal business document, memo, report, meeting record, plan, or vendor communication. Do NOT invoke for RFP / 招標規格 / 需求規格書 (use rfp-writing), for pure language cleanup with no structural work (use avoid-ai-writing / clean-ai-writing), or for casual chat, creative writing, marketing copy, or code comments.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
7.7 KB, as published. Nobody here has run it
Formal Internal Document Structure
Produce internal business documents that help a reader decide, approve, assign work, review risk, brief management, communicate with a vendor, or verify completion. The output is a usable draft — complete sections that paste straight into a document — not a critique.
Output Language
Match the language of the user's request, and apply it to all user-facing output — option labels, generated-document headings, table column names — not just prose. If the user explicitly asks for another language, that wins.
Language follows the request, not the source material. When the user writes in Chinese but the uploaded document, code, or reference is in English, output stays Chinese.
If the request is in Chinese, use Traditional Chinese (Taiwan business usage) and keep established technical terms in English.
The English in this file is structural labelling for you, not literal output. Never mirror this file's language into your response.
When to use
Use when the request names one of these artifacts (or a clear equivalent):
- 簽呈與核准文件, 經費 / 人力 / 資源申請
- 會議紀錄 (內部、廠商、審查)
- 專案規劃 / 執行計畫, 組織與人力規劃, 訓練與培育文件
- 需求文件, 評估報告, 風險說明
- 採購與廠商溝通文件 (詢問、澄清、合作範圍)
- 制度 / 流程 / 作業規範, 問題追蹤與改善報告, 驗收與交付文件
- 對主管簡報文字, 對外回覆稿, 跨單位協調材料
When NOT to use
Each of these has conventions that conflict with internal-business-doc rules:
- RFP / 招標規格 / 需求規格書 / 投標 issuer specs → use
rfp-writing. That skill governs documents sent out for bidding; this one governs documents that circulate inside the organization. - Pure language cleanup (remove AI-isms from existing text, no structural change) → use
avoid-ai-writing/clean-ai-writing. Reach for this skill when the job also involves organizing a document by type. - Casual conversation, creative writing, marketing / advertising copy, or code comments — unless the user explicitly asks for an internal-document style.
Do not force every document into a project-plan shape. Pick the structure from the reader's need (see below).
Core writing principles
-
Write like an experienced internal staff member — a project owner, IT lead, HR partner, procurement officer, or risk reviewer preparing the document for internal circulation. Grounded, practical, business tone. Not promotional, not slogan-like.
-
Every sentence must earn its place. Keep a sentence only if it adds a decision, requirement, constraint, risk, responsibility, deliverable, timeline, verification method, follow-up action, or concrete explanation. Cut filler, restatement, and abstract conclusions. This is the same discipline
avoid-ai-writingenforces, applied while you build the document. See references/language.md for the prohibited-phrase list and fixes. -
Use affirmative planning language. Write what will be done, proposed, provided, and how it is verified — 預計採…, 規劃…, 建議…, 應提供…, 需完成…, 由…負責…, 成果包含…, 後續依…辦理…. Avoid negative framing (不建議…, 不宜…, 不是…而是…); when contrast is needed, rewrite it as a direct implementation statement.
Poor: 不建議僅以會議討論作為結論,而是要形成後續追蹤項目。 Better: 會議結論需整理為後續追蹤項目,並列明負責單位、預計完成時間及檢核方式。
Pick the structure from the reader's need
Choose by what the reader must decide, approve, understand, or execute. Full templates in references/structures.md:
| Reader need | Document type | Template |
|---|---|---|
| Sign off / approve / fund | 決策或核准文件 (簽呈, 採購, 預算) | A |
| Compare options / assess risk | 評估報告 | B |
| Record what was decided | 會議紀錄 | C |
| Execute a plan | 執行 / 作業計畫 (專案, 訓練, 流程) | D |
| Coordinate with a vendor | 廠商 / 對外溝通 | E |
Then apply three structural habits (detail in references):
- Keep chapters concentrated — group related ideas (背景+現況, 範圍+假設, 交付物+驗收, 風險+控管). Fewer, stronger chapters beat many thin ones.
- Summary before detail — open a heavy section with one or two sentences of planning intent / conclusion, then the steps or tables.
- Don't repeat other documents — in an appendix or follow-up, reference the parent (本文件承接主文件之…) and focus only on this document's unique purpose.
Content must be concrete enough to execute
Stop short of conceptual statements. For each important activity, name the responsible unit, owner, timeframe, scope boundary, deliverables, acceptance criteria, review cadence, and follow-up mechanism. Every activity gets a corresponding output (meeting→紀錄/決議/待辦; review→評估備忘/發現/追蹤清單; vendor work→交付物/驗收紀錄/問題log; system task→測試/部署/rollback 紀錄).
Poor: 需加強後續管理。 Better: 後續由承辦單位每月彙整執行情形,內容包含進度、待辦事項、風險、需主管協調事項及預計完成時間。
Drafting patterns for each paragraph type (purpose, background, execution model, responsibility split, vendor scope, risk/control, acceptance) are in references/language.md.
Revision checklist
Run before finalizing:
- Does every section have a clear purpose, and can any two be merged?
- Is any paragraph repeating a point made elsewhere?
- Are tables necessary and decision-oriented (not restating the prose)?
- Are responsibilities — responsible / supporting / review / vendor / escalation — clearly assigned?
- Are time, deliverables, and acceptance criteria specific enough to act on?
- Does the document connect the proposal to daily work (who owns it, what outputs, how reviewed)?
- Are vendor requirements concrete enough to discuss or quote against?
- Are risks, assumptions, and constraints stated plainly?
- Are all AI-style slogans and filler phrases removed (see references/language.md)?
- Is the tone formal, practical, and fit for internal review?
De-AI finishing pass
After the checklist passes, offer avoid-ai-writing-zh as a deep de-AI pass over the draft, a far richer AI-ism ruleset than item 9's own light check (references/language.md). Run it only if the user agrees; otherwise finalize as-is. This is the last step before final output.