Documentation
Technical documentation patterns, structure, maintenance, and avoiding common documentation failures.
Iván
@ivangdavila
What This Skill Does
Provides structured patterns for writing, organizing, and maintaining technical documentation. Covers README essentials, API docs, code examples, versioning, and common failure modes.
Replaces ad-hoc documentation practices by offering a proven hierarchy (README → Getting Started → Guides → Reference → Troubleshooting) and maintenance rules that prevent outdated content.
When to Use It
- Write a README that gets users to first success in under 5 minutes
- Structure API documentation with complete request/response examples and error codes
- Add runnable code examples that are tested in CI to prevent rot
- Create a versioned documentation system with migration guides between major versions
- Audit existing docs for common failures like missing context, wall of text, or broken links
- Set up a maintenance workflow with review dates and automated link checking
Install
$ openclaw skills install @ivangdavila/documentationStructure Hierarchy
- README: what it is, how to install, quick example — 5 minutes to first success
- Getting Started: guided tutorial for beginners — one complete workflow
- Guides: task-oriented ("How to X") — goal-focused, not feature-focused
- Reference: exhaustive API/CLI docs — complete but not for learning
- Troubleshooting: common errors with solutions — search-optimized
README Essentials
- One-sentence description — what problem it solves
- Installation — copy-paste command that works
- Quick start — minimal example that actually runs
- Link to full docs — don't cram everything in README
Missing any of these = users bounce before trying.
Code Examples
- Every example must be tested — untested examples rot within months
- Show complete runnable code, not fragments — users copy-paste
- Include expected output — confirms they did it right
- Bad:
client.query(...)/ Good: full script with imports, setup, and output - Version-pin examples:
npm install package@2.1.0notnpm install package
API Documentation
- Every endpoint needs: method, path, parameters, request body, response, error codes
- Show real request/response bodies — not just schemas
- Include authentication in every example — most common missing piece
- Document rate limits and pagination upfront — not buried in footnotes
- Error responses need as much detail as success responses
What Gets Outdated
- Screenshots — UI changes, screenshots don't
- Version numbers — hardcoded versions become wrong
- Links — external sites move, break constantly
- "Current" anything — write timelessly or add review dates
- Feature flags and experimental warnings — often forgotten after GA
Maintenance Patterns
- Docs live next to code — same repo, same PR. Separate repos drift
- CI checks for broken links —
markdown-link-checkor equivalent - Runnable examples as tests — if example breaks, build fails
- Review date in docs: "Last verified: 2024-01" — signals freshness
- Delete aggressively — outdated docs worse than no docs
Common Failures
- Documenting implementation, not usage — users don't care how it works internally
- Assuming context — define acronyms, link prerequisites
- Wall of text — use headings, bullets, code blocks liberally
- "See X for more info" without link — friction kills follow-through
- Changelog as documentation — changes ≠ how to use current version
Writing Style
- Imperative mood: "Run the command" not "You can run the command"
- Second person: "you" not "the user"
- Present tense: "This returns X" not "This will return X"
- Short sentences — one idea per sentence
- Active voice: "The function returns X" not "X is returned by the function"
Searchability
- Use words users search for — not internal jargon
- Error messages verbatim in troubleshooting — users paste exact errors
- Multiple ways to describe same thing — alias common variations
- H2/H3 headings are SEO — match user queries
- Avoid clever titles — "Getting Started" beats "Your Journey Begins"
Versioned Documentation
- Major versions need separate docs — v1 users shouldn't see v2 docs
- Migration guides between versions — step-by-step, not just changelog
- Default to latest stable, link to older versions
- Mark deprecated features clearly — don't just remove
- URL structure:
/docs/v2/not query params
README Anti-patterns
- Badge spam — 15 badges before content
- Massive feature lists — save for marketing page
- No installation instructions — assuming everyone knows
- Screenshots without context — what am I looking at?
- License-only README — legal compliance ≠ documentation
Top skills in this category
self-improving agent
@pskoettCaptures and maintains learnings, errors, and corrections. Use when a command fails, the user corrects an assumption, a capability is missing, knowledge is outdated, or a better approach is found. Review relevant saved guidance before major tasks, reuse, and promotion; revalidate it according to wha
Self-Improving + Proactive Agent
@ivangdavilaSelf-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
ontology
@oswalpalashTyped knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
Humanizer
@biostartechnologyRemove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.
Obsidian
@steipeteWork with Obsidian vaults (plain Markdown notes) and automate via obsidian-cli.