Hubbestaiskill.com

Home / All Skills / Workflow & Core Tools / technical-writing

Workflow & Core Tools · cursor/plugins

technical-writing

Layered technical-writing standard: Diátaxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use for /technical-writing or when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit

Author cursorGitHub cursor/plugins ↗Stars 7221License 仓库根无统一 LICENSE（GitHub API license 为 null）；该 skill 随 pstack 插件分发，插件自带 pstack/LICENSE（MIT，Copyright (c) 2026 Lauren Tan）Commit df3fb154fb

Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Stasks four writing standards with one question per layer: Document type (Diátaxis) → Audience-oriented sentences (Google developer style) → Single-sentence payload (STE) → Disambiguation (Global English), progressively narrowing down to 'instantly clear at first read'.

pstack/skills/technical-writing/SKILL.md

The goal is writing a tired engineer understands on the first read. Four layers get you there, one question each: what kind of document is this, how do sentences address the reader, how much does each sentence carry, and can any sentence be read two ways. Apply all four.

The four layers are not a parallel checklist but progressive constraints: first determine the document type, then manage sentence person/voice, then cap the information payload per sentence, and finally eliminate ambiguous readings.

Establishes three overriding general rules atop the layers: delete functionless words, use short and common words, and when a rule degrades a sentence, rewrite the sentence rather than stubbornly enforcing the rule—rules serve the reader rather than sounding machine-written.

pstack/skills/technical-writing/SKILL.md

- **Cut every word that does no work.** If the sentence survives without a word, the word goes. "In order to" is "to". "It is important to note that" is nothing.

The third rule in the same section, 'When a rule makes a sentence worse, fix the sentence another way or leave it alone…A sentence that follows every rule and sounds like a machine wrote it has failed,' serves as the anti-dogmatic safety valve of this standard.

Diátaxis mode triage: Classify every document into one of tutorial/how-to/reference/explanation using two questions: 'Guiding action or building understanding × Serving learning or working'; mixing modes is forbidden, and cross-mode content must be split via links.

pstack/skills/technical-writing/SKILL.md

One document, one mode. Two questions pick it: does the content inform action (doing) or understanding (thinking), and does it serve learning or work?

Provides colloquial behavioral guidelines for each mode (tutorial = 'we' + imperative + early visible results; how-to = practical steps only; reference = purely descriptive without persuasion; explanation = the only place where opinions are allowed), culminating in 'Don't mix modes: no reference tables inside a tutorial…Split and link instead.'

The sentence-level rules directly incorporate three external standards with their sources and fetch dates recorded in the body: Google developer style (developers.google.com/style), STE controlled language (asd-ste100.org Issue 9 2025), and Global English (Kohl, SAS Press—transcribed via Internet Archive and SAS sample chapters); all fetched at writing time, not via runtime calls.

pstack/skills/technical-writing/SKILL.md

Source: diataxis.fr, fetched 2026-07-18.

Each of the four sources includes a 'Source: … fetched 2026-07-18.' line (diataxis.fr, developers.google.com/style, asd-ste100.org, Kohl's book); STE self-discloses that 'The numbered rules and dictionary live in the spec PDF. The principles above are the transferable core.'—honestly stating it only transcribes the transferable core.

Repository vocabulary discipline: 'The codebase is the vocabulary'—write actual symbol/file/command names and forbid invented jargon or metaphors; when a new prohibited word is identified, do not directly edit unslop, but instead propose the 'prohibited word → replacement' alongside a diff in the response.

pstack/skills/technical-writing/SKILL.md

The codebase is the word list. Write the real symbol, file, flag, or command name, not a synonym or a description of it.

Clear collaboration boundaries: 'Propose a new offender and its replacement as an addition to `unslop`'s abstract-metaphor rule in your reply, with the diff. Don't edit that skill.'—the vocabulary is owned by unslop; this skill only makes proposals.

Hard-binds with unslop and expands coverage: any document touched by this skill must pass through unslop (a catalog of AI tone, filler words, and formatting tells); PR descriptions and commit messages are also writing targets (all layers apply except Diátaxis).

pstack/skills/technical-writing/SKILL.md

- Apply the **unslop** skill to every doc this skill touches. That skill owns the slop-pattern catalog: AI vocabulary, filler, hedging, formatting tells.

Expands 'document quality' from body text to engineering artifacts: 'PR descriptions and commit messages are writing too…A PR body is a briefing that a reviewer can read in under a minute.'

Provides worked examples (Before/After rewriting demonstrations) and an 8-item review checklist to operationalize abstract standards into directly actionable acceptance questions; declarative rules require that 'numbers, paths, and counts must be true against the landing commit, accompanied by reproducible commands.'

pstack/skills/technical-writing/SKILL.md

Make every count or tree claim true at the commit that lands it, and include the command that regenerates it.

The Before/After example rewrites a configuration guide packed with passive voice and fluff into three statements containing actual commands; the Review checklist contains 8 items (single mode / imperative mood / one meaning per sentence / deletable words / referential ambiguity / one noun per entity / colloquial words / accurate symbols and counts).

2Core Capabilities

01Document type triage: Diátaxis single mode selection + forbidden mode mixing + cross-mode linking

02Sentence-level rules: you/present tense, clear agent, imperative mood, preposed conditions, prohibiting 'simply/easy/quickly/please', etc.

03STE controlled language: one meaning per sentence, splitting overly long sentences, preserving 'the'/'a' for disambiguation, one meaning per word, direct imperative for procedures, minimizing -ing forms

04Global English disambiguation: adjacent only/not, breaking noun strings, unambiguous reference, retaining structural particles, prohibiting slashes and abbreviations, one noun per entity

05Pacing and perspective control: alternating sentence lengths, one meaning per sentence ≠ one length per sentence, allowing perspectives where permitted by mode, concrete over sterile

06Discipline for authentic symbols + unslop linkage + PR/commit coverage + worked examples + 8-item review checklist

3External Dependencies

Kind	Dependency / Endpoint
network	内容来源引用（撰写期抓取的知识来源，非运行时调用）：diataxis.fr
network	内容来源引用（非运行时调用）：developers.google.com/style
network	内容来源引用（非运行时调用）：asd-ste100.org（STE 规范 Issue 9, 2025）
network	内容来源引用（非运行时调用）：Kohl《The Global English Style Guide》（SAS Press，经 Internet Archive/SAS 样章转写）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Can be auto-loaded

For Administrators

Register source and version

Evidence & Triggers (File Paths & Excerpts)

SKILL.md: Pure prompt instructions and rule text, without script execution or external impact

Audit Notes: Pure prompt-based instruction skill: no script execution, no outbound network requests, no credential reading, and no local file writes, with zero I/O at runtime (all four layers of standards and examples are inlined in SKILL.md); additionally, disable-model-invocation: true ensures it is only triggered explicitly by the user (/technical-writing or during writing/reviewing tasks). The only considerations are stylistic risks from mechanistic rule adherence and the freshness of static knowledge sources, posing no code-level risks.

5Second-Pass Verification

[ok] 单文件零资产 — 技能目录仅 SKILL.md（含隐藏文件 glob 确认），无 references/scripts/data。
[ok] 四层标准与 description 对应 — 正文标题/内容与 description 四层（Diátaxis structure / Google developer style sentences / STE instruction rules / Global English syntax）一一对应；每层均给独立规则块与来源行。
[ok] 来源引用为静态文字、非运行时调用 — 4 处 'Source: … fetched 2026-07-18.' 均为引用标注；无 urlopen/curl/requests 等运行时拉取代码。
[ok] 规则内容与所引规范一致（抽样） — Google style 段 'Talk to the reader as "you", in the present tense.'、STE 段 'Split instructions longer than about 20 words and other sentences longer than about 25.'、Global English 段 'Break up long noun strings' 等与规范主旨相符；并如实声明 STE 只转写可迁移核心、完整规则在 PDF。
[ok] unslop 联动指向真实资产 — pstack/skills/unslop/SKILL.md 存在（glob 确认）；SKILL.md 明确 'That skill owns the slop-pattern catalog'——职责划分与仓库结构一致。
[ok] 无运行时 IO/凭证/写入 — token 扫描仅命中来源标注行；全文无 shell/网络/文件命令。
[ok] 夸大检查 — 'writing a tired engineer understands on the first read' 是质量标准陈述而非可测功能承诺；worked example 与 checklist 均可直接执行，无越界承诺。
[ok] license/stars/commit 元数据 — GitHub API：cursor/plugins stars 7221、pushed_at 2026-09-09T08:18:30Z、license null（仓库根）；pstack/LICENSE 为 MIT（Copyright (c) 2026 Lauren Tan）；本地 git HEAD == pin df3fb154fb982fb83f649de8646d4af6a0cb16b3。

6Verdict & Scenarios

Consolidates four authoritative writing specifications into a single, progressively executable standard, providing a 'one-sentence question' for each layer to lower the barrier to application.

Transparent sources: every specification includes provenance and fetch dates for verifiable auditing; transparently declares that only the transferable core of STE is covered when full transcription is infeasible.

Anti-dogmatic safety valve (rewrite the sentence if a rule degrades it) ensures the standard applies to real-world writing rather than merely passing a linter.

Comprehensive coverage of engineering artifacts: PRs and commits are incorporated, which directly boosts code review text quality when combined with unslop vocabularies and 'counts must be true at commit'.

Zero I/O and zero dependencies, allowing seamless migration of the entire directory to any host; Before/After examples translate abstract standards into concrete, imitable formats.

Best for: Engineering teams and individuals who treat documentation, RFCs, READMEs, PR descriptions, and commit messages as first-class deliverables; skill authors seeking to unify multiple writing standards and reference its 'layered + sourced + example-driven' structure; sessions integrating into existing coding disciplines (the unslop/poteto-mode family).

Not for: Scenarios requiring automated rewriting or batch-polishing pipelines (it is a pure standard without execution capability and must be applied manually by the agent); scenarios seeking marketing or creative copywriting (standards target technical documentation, with opinions permitted only in explanation mode); scenarios requiring real-time network retrieval for the latest specification versions (knowledge consists of static snapshots).

Install Agent-Ready

① Verified Mirror 2026-09-09

Option A · Direct DownloadDownload technical-writing.tar.gz

sha256: bf0a9d23048af4b7…

Option B · Copy Guide for Agent

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit df3fb154fb.