technical-writing

SkillDocs & knowledge

Helps your agent write clearer docs, READMEs, and API references using a technical writing skill.

Available today. Use it from your connected AI after setup.

Add ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Then ask your AI: use the technical-writing skill

About this skill

Patterns for clear technical writing in documentation, READMEs, API references, and design documents.

What this skill tells your AI

The instructions your AI receives, as published by npc-worldwide/npcsh in skills/technical-writing/SKILL.md and read by ahel’s review.

Patterns for clear technical writing in documentation, READMEs, API references, and design documents.

Clarity

One idea per sentence. One topic per paragraph.

  • Use simple subject-verb-object order. Avoid nested clauses.
  • Replace jargon with plain language unless the term is standard for the audience.
  • Define acronyms on first use. Do not assume the reader knows internal project names.
  • Use active voice: "The scheduler retries failed tasks" not "Failed tasks are retried by the scheduler."

Precision

Be exact about behavior, not suggestive.

  • Write "Returns a list of strings" not "Returns some data."
  • Write "Throws ValueError if the input is negative" not "May throw an error for bad input."
  • Write "Timeout is 30 seconds" not "Timeout is fairly short."
  • Distinguish "must" (requirement), "should" (recommendation), and "may" (optional).

Structure

Documents follow a predictable hierarchy:

  • What this is (one sentence).
  • What problem it solves (one paragraph).
  • How to use it (minimal working example).
  • Configuration or parameters (table or list with types and defaults).
  • Error handling and edge cases.
  • Limitations and assumptions.
  • Links to related docs or source code. Do not bury the example under three paragraphs of motivation.

Code-Examples

Every concept must have a minimal reproducible example.

  • Show the simplest possible call that produces a valid result.
  • Show one error case if the failure mode is non-obvious.
  • Include expected output as a comment or separate block.
  • Keep examples copy-pasteable. No pseudocode in API docs.

Cross-References

Never leave the reader hunting for related information.

  • Link to prerequisite concepts before using them.
  • Link to deeper dives from high-level summaries.
  • If a term is defined elsewhere, link the first use in each major section.
  • Keep links working. Broken links erode trust faster than missing links.

Revision-Checklist

Before marking a doc complete:

  • Can a new reader follow the example without asking for help?
  • Are all claims verifiable (traceable to code, test, or source)?
  • Are there any ambiguous pronouns ("it", "this") that could mislead?
  • Is every number, parameter, and default current with the code?
  • Does the doc answer "what happens if this goes wrong?"

Signals

GitHub stars
482
Forks
32
Last commit
Sep 2026
Advanced
Item type
skill
Key
technical-writing-npc-worldwide
Source
github.com/npc-worldwide/npcsh