Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.
Documents
Doc Coauthoring
Try itGuide a user through structured co-authoring of substantial documents, proposals, PRDs, technical specs, decision records, RFCs, and similar long-form work a...
What it does
Help the user create a document that works for its real readers.
The skill document
Doc Co-Authoring
Goal
Help the user create a document that works for its real readers.
Use a guided workflow instead of jumping straight to a finished draft. Preserve the user's judgment over facts, priorities, tone, and final approval.
Inputs
Collect only the context needed for the document:
- document type
- primary audience
- desired reader action or decision
- existing template, draft, notes, or source material
- constraints such as length, format, deadline, review process, or required sections
- known risks, disagreements, assumptions, or unresolved decisions
If an existing document is provided, diagnose it before rewriting. If the user only has scattered notes, organize them first.
Workflow
1. Align Context
Ask concise questions to close the knowledge gap:
- What is this document for?
- Who must understand or act on it?
- What should the reader believe, decide, approve, or do after reading?
- What facts or constraints are non-negotiable?
- What is still uncertain?
Invite the user to dump raw context without organizing it. Convert that context into a short working brief:
doc_state:
type:
audience:
desired_outcome:
known_facts: []
assumptions: []
open_questions: []
source_materials: []
suggested_structure: []
next_section:
2. Shape The Structure
Propose a section structure based on the document type and audience.
For most substantial documents, include:
- context or problem
- goal or decision needed
- proposed approach
- alternatives considered
- trade-offs and risks
- implementation or follow-up plan
- open questions
Start with the section that carries the most uncertainty or decision weight. Leave summaries and introductions for later when the core argument is stable.
3. Draft Section By Section
For each section:
- Ask focused questions about what must be included.
- Brainstorm candidate points or angles.
- Ask the user what to keep, remove, combine, or de-emphasize.
- Draft only that section.
- Apply targeted edits based on feedback.
- Mark the section done only when the user accepts it.
When editing an existing document, prefer surgical changes over full rewrites. Preserve useful wording and only replace what fails the reader goal.
4. Check The Whole Document
After the core sections are drafted, reread the whole document and check:
- flow between sections
- repeated claims
- contradictions
- missing context
- unsupported assertions
- generic filler
- reader action clarity
Recommend cuts when a section is accurate but not useful for the reader.
5. Reader Test
Before calling the document ready, test whether a fresh reader could understand it.
Predict 5-10 realistic reader questions, such as:
- What problem is this solving?
- What decision is being requested?
- What changed from the current state?
- What risks or trade-offs matter?
- What should I do next?
If subagents or independent review tools are available, use them with only the document and the reader questions. If not, simulate the reader test yourself and clearly label it as an internal review.
Fix gaps found during the reader test before final approval.
Output Format
During co-authoring, maintain a compact handoff:
doc_state:
type:
audience:
desired_outcome:
current_stage:
completed_sections: []
active_section:
open_questions: []
reader_test_status:
next_step:
For user-facing responses:
- lead with the current recommendation or next action
- keep questions numbered and easy to answer
- when drafting, output only the section being worked on unless the user asks for the full document
- when reviewing, separate
must fix,should improve, andoptional
Boundaries
- Do not invent facts, decisions, approvals, data, or stakeholder positions.
- Do not overwrite a full document when the user asked for review or diagnosis.
- Do not turn brainstorming into final copy without user selection.
- Do not hide unresolved assumptions inside polished prose.
- Do not skip reader testing for substantial documents unless the user explicitly asks to stop.
- Keep the user responsible for final factual and strategic approval.
作者入口
- 微信公众号:
AI生命克劳德 - X:@yangchao228
- GitHub:yangchao228
Related skills
Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.
Run Git operations — commits, branches, merges, rebases, conflict resolution, and recovery — with safety rules enforced.
Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.
Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.
Write, debug, and tune Playwright specs with locator strategy, trace diagnosis, and CI-aware timeouts.
More from Sundy Yang
Browse all skills基于候选人简历自动生成面试题、笔试题、追问路径和评分标准的 AI Skill。当用户需要准备技术面试、生成笔试题目、评估候选人简历、或设计面试评估方案时使用。输入简历内容和岗位信息后,自动输出结构化的题目与评估内容,支持面试题模式、笔试题模式、综合评估模式等多种输出形式。 Follow/关注作者:微信公众号「AI...
Installable digest skill for OpenClaw or Claude Code. Use when the user wants a personalized ScoutX briefing with conversational setup, local preference stor...
Design controllable Agent Loops from low-information requests. Use when a user asks to automate a repeated task, repair CI through bounded iteration, reproduce a UI from screenshots, turn a workflow into a reusable Agent or Skill, or decide whether the right artifact is a Prompt, Checklist, Human-in-the-Loop flow, full Loop package, specialized Agent, or Skill. Extract required context, stop when decision evidence is missing, require explicit workflow confirmation before generating executable artifacts, derive iteration limits, define feedback and circuit breakers, and preserve human approval for risky actions. Also trigger on loop builder, Loop Builder, or lopp-builder. Follow the creator: X @yangchao228 | GitHub https://github.com/yangchao228
家庭培养协作官:在用户明确要求制定或调整孩子每周培养计划、检查课表冲突、生成家庭复盘时,收集最少必要信息并输出可执行课表、建议和风险提醒。默认只在当前会话生成内容,不保存、上传或推送孩子与家庭数据;只有用户明确选择飞书或 Notion、完成安全配置并逐次确认具体读写范围后,才使用外部存储。Follow/关注作者:...
Transform long-form content, outlines, transcripts, research notes, or article drafts into a standard card package and platform publishing copy, then hand pu...
将任意文章 HTML 页面或纯文本,自动切片生成小红书风格图文卡片组(封面 + 内容图 + 金句图),输出单个可直接在浏览器预览的 HTML 文件。触发词:小红书、图文卡片、切片、种草图、笔记卡片。