technical-writing

Applies plain-language rules to docs, comments, commit messages, and error messages.

15|1|Updated Jan 3, 2026
One-click install
npx skills add https://github.com/fprochazka/claude-code-plugins --skill technical-writing-fprochazka
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: technical-writing
Source: https://github.com/fprochazka/claude-code-plugins/tree/main/plugins/prose/skills/technical-writing
Command: npx skills add https://github.com/fprochazka/claude-code-plugins --skill technical-writing-fprochazka

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Technical text often restates code, drifts into marketing language, or sounds machine-generated, making it stale and hard to read. This Skill enforces a rule set that keeps documentation, comments, and messages focused on what the code cannot say. ## Core Features & Use Cases - Content rules: Write down why, gotchas, business meaning, and non-obvious relationships; delete restatements of code, change narration, and ticket references. - Form rules: Short active sentences, plain verbs, one name per thing, no marketing adjectives or LLM vocabulary, derived from ASD-STE100 Simplified Technical English. - Per-artifact guidance: Specific conventions for doc pages, READMEs, doc comments, inline comments, commit messages, MR/PR descriptions, error messages, release notes, and chat or ticket comments. - Use Case: When asked to write an MR description, produce a reviewer-focused structure: why with ticket link, what changed, how to review, how to release, and risks. ## Quick Start Use the technical-writing skill to rewrite this README section so it is plain, active, and free of filler.

Frequently Asked Questions about technical-writing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write better doc comments and Javadoc?▼

Make the first sentence stand alone as the summary, state what the caller must know such as contracts, side effects, and thread-safety, and never restate the signature. Follow the language convention for the summary verb, and skip comments on getters, setters, and obvious constructors.

What should a good MR or PR description contain?▼

Structure it for the reviewer: why with the ticket link, a short list of what changed, how to review including where to start and how it was tested, how to release with migrations and flags, then risks and follow-ups. Omit anything the diff already shows.

When should I not use these writing rules?▼

Never apply them to code, identifiers, or chat replies to the user, which belong to a separate reply-style rule set. The rules target text people read, such as docs, comments, commit bodies, and error messages.

How do I make AI-generated text sound less like AI?▼

Delete marketing adjectives and LLM vocabulary like leverage, delve, and streamline, replace metaphors with literal phrases, and avoid rule-of-three padding and em-dash pile-ups. Run the self-lint checklist: split long sentences, activate passive voice, and use plain verbs.

What belongs in an inline code comment?▼

Only what the code cannot say: intent, constraints, trade-offs, or a rejected simpler approach and why. If someone could write the comment from the code alone, delete it, and never put ticket references in code comments.