AGENT SKILL

technical-writing

Writes and reviews project copy, READMEs, developer documentation, tutorials, API and CLI references, UI text, errors, and agent instructions. Use whenever creating or revising project-facing prose; grounds claims in current behavior and preserves meaning while improving clarity.

License 许可
MIT
最近更新
2026-11-12
Stars
★ 1
Forks
0

01 前言 Front Matter

name
technical-writing
metadata
{ "skill-graph-id": "plugin/10x-swe/technical-writing", "skill-graph-uses": "", "skill-graph-used-by": "plugin/10x-swe/brainstorming" }
description
Writes and reviews project copy, READMEs, developer documentation, tutorials, API and CLI references, UI text, errors, and agent instructions. Use whenever creating or revising project-facing prose; grounds claims in current behavior and preserves meaning while improving clarity.

02 资源文件 8

03 SKILL.md 正文

<objective> Help a reader understand a product, choose an action, and complete it correctly. Write concise, accurate prose for humans and agents. Follow project terminology, voice and documentation policy before general style preferences. </objective>

<quick_start> Identify the reader, their goal, the artifact and the first useful outcome. Inspect the current capability contract before drafting. For a small edit, revise and check the affected sentence; for a substantial rewrite, follow the process. Read README guidance for landing pages, developer documentation for procedures/references, and language and fidelity for wording and copy reviews. </quick_start>

<principles>

  • Lead with purpose, usefulness or the reader's next action. Define unfamiliar concepts with a concrete example.
  • Use familiar words, active verbs and consistent names. State the actor when responsibility matters. Keep one topic per paragraph.
  • Preserve facts, conditions, scope, uncertainty, negation and requirement strength. Keep code, flags, identifiers, paths, units, quotations and URLs exact.
  • Use a fact brief as evidence, not copy to paste. Express verified scope positively instead of filling product copy with absence lists or repeated disclaimers.
  • Describe implemented behavior precisely. Skills prescribe work; they do not guarantee agent outcomes. Scope local/offline, safety and performance claims to the verified capability.
  • Use stricter sentence discipline for procedures, errors and agent instructions. Keep README explanations natural; preserve approved product voice in UI and marketing copy.
  • Keep current capabilities, setup, contracts and operations in repo docs. Put plans, TODOs, experiments and detailed validation history in the project wiki/archive. Retain essential requirements and recovery steps with the code. </principles>

<process>

  1. Ground. Read the project instructions and relevant architecture. Build a compact fact brief: audience, task, current capabilities, prerequisites, exact examples and important effects. Verify uncertain commands with source/help. Keep research notes separate from the published copy.
  2. Structure. Give the artifact one job: README orientation/first success; how-to task completion; reference exact contracts; architecture current boundaries; UI copy the user's decision or next step. Load only the relevant reference.
  3. Draft. Write the shortest complete explanation for that reader. Put prerequisites before actions. Show a concrete useful task and its expected result where applicable. Link deeper detail instead of repeating it.
  4. Challenge. Compare every consequential claim and example with the fact brief. Check modality, actors, conditions, errors, defaults, privacy, data ownership and approval. Check what each condition modifies; preserve unconditional constraints across all actions. Fix misleading implications as well as literal errors.
  5. Edit. Remove repetition, vague promises, jargon, filler and development narration. Preserve necessary detail even when a sentence must be longer. Apply language and fidelity.
  6. Verify. Check links, anchors, formatting and example syntax. Run the safe representative task when authorized; syntax checks alone do not establish that a workflow works. Review changed UI copy in its actual context when practical. Report check limits outside the document. </process>

<collaboration> Use one agent for routine edits. For substantial rewrites, an investigator can supply verified facts, a writer draft independently, and a different reviewer challenge the claims before an editor combines the best parts. A fresh reader can attempt the first task. See review and agent roles. Choose models by task needs and observed results, not fixed brand rankings. Delegation, external model calls and publication require existing authorization; this skill grants none of them. Share only material allowed for that destination. </collaboration>

<success_criteria> The intended reader can explain the purpose and find the next useful action. Consequential claims match current behavior; requirements and examples agree. Rewrites preserve meaning and operational safeguards. Links/examples have the appropriate checks. Deliver the requested artifact without editorial narration; put rationale, evidence and unresolved questions in a separate review when useful. Source attribution and adaptation details are in sources. </success_criteria>

04 同仓库更多 Skills

cercova-studios
cercova-studios/10x-swe

turnstile-spin

Set up Cloudflare Turnstile end-to-end in a project. Scan the codebase, create the widget via the Cloudflare API, embed it where user requests need bot verification (form submissions, SPA actions, API endpoints, download links, comment or vote submissions, etc.), wire canonical server-side siteverify in the customer's existing backend, validate, and persist the skill. Load this when a user asks to add Turnstile, set up CAPTCHA, protect a form or endpoint from bots, or fix a Turnstile integration. Uses Cloudflare's documented Turnstile contract.

后端与接口
★ 1
cercova-studios
cercova-studios/10x-swe

frontend-design

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.

设计师前端开发视觉设计
★ 1
cercova-studios
cercova-studios/10x-swe

source-insights

Prepare inspected source evidence, preserve exact citations, and publish reviewed insights into an explicit local library.

★ 1
cercova-studios
cercova-studios/10x-swe

skill-graph

Discover skills and workflows through independent lexical search, relationship paths and a terminal graph.

Skill与智能体开发
★ 1
cercova-studios
cercova-studios/10x-swe

session-search

Recall public session turns with exact provenance, bounded excerpts and independent piped exports.

★ 1
cercova-studios
cercova-studios/10x-swe

newsletter

Produce a private local AI newsletter from bounded public research, with cited deep dives, HTML/Markdown output, Dagu scheduling and seven-day retention.

邮件与沟通资料检索与研究
★ 1