agentleFS
oqtopus-teamoqtopus-team/qdash/.agents/skills/technical-docs/SKILL.md

technical-docs

Skill54 starsChanged 24 days agoApache-2.0View on GitHub

Write, rewrite, and review pages under docs/ in the QDash repository. Use whenever a task creates or edits documentation: a user-guide procedure, a developer-guide page, a reference table, the docs sidebar, or a review of an existing page. Applies the repository's two-reader policy (docs/developer-guide/docs-guidelines.md), the Google Developer Documentation Style Guide, and Diataxis.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. getsentry/sentry-docs★ 449same name
    2. poorvith-mp/skills-writing★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Diagrams for relationships that prose hides: Use a small Mermaid flowchart, sequence diagram, state diagram, or class diagram when the …

    line 57
  2. Distinctive:

    Do not diagram a list, a single linear action, or the same information already clear in a table.

    line 64
  3. Distinctive:

    Check every stated behaviour (default values, limits, orderings, fallbacks, error handling) against the implementation or its tests, not …

    line 73
  4. Very specific:

    Run bun run docs:build in docs/ when links or the sidebar changed.

    line 80
  5. Distinctive:

    When a check fails, fix the page, not the code, unless the task is about the code.

    line 83
  6. Distinctive:

    Read the page once as its reader, with these questions, and rewrite until every answer is yes:

    line 87

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Technical Documentation

docs/developer-guide/docs-guidelines.md is the rulebook: readers, sections, layout, what every page must and must not contain. Read it first and do not restate it here. This skill is the working method: how to choose the page type, how to write the sentences, how to check them against the code, and how to review before finishing.

Pick the reader and the type

Decide both before writing a line, and keep one page to one type.

Reader's goalDiataxis typeQDash sectionShape
Get a job done in the applicationHow-todocs/user-guide/When to use it, then numbered steps; names of pages and buttons in bold; no internals
Set up or operate a deployment for the first timeTutorialdocs/user-guide/setup.md, operations.mdA path that works end to end; commands the reader can paste
Understand how a part works or why it is built this wayExplanationdocs/developer-guide/Responsibility boundary, mechanism, invariants with reasons, how to extend and test, file list
Look up a factReferencedocs/reference/Flat, exhaustive tables; no narrative

A topic that needs both a procedure and a mechanism gets two pages, one per section, linked in one line each way (How this works: [...] on the user page). When a page mixes types, move the foreign paragraphs to the page of the other type rather than smoothing them in place.

Write

Google style, applied to this repository:

  • Lead with the answer. The first sentence of a page or section carries its main fact. No "This document describes", no history of how the feature came to be.
  • One idea per sentence, active voice, present tense. "The API derives the reference on every read", not "the reference will be derived by the API".
  • Define before use. Introduce a term (record, run, reference, planner) with its meaning the first time; afterwards use exactly that word. Do not rotate synonyms for the same thing.
  • Concrete over abstract. A rule shown on one real input (Short T1 (<20 μs) to SHORT_T1) beats a paragraph of generalities. Code and command samples must be runnable or taken verbatim from the code.
  • Name what the reader will open. Developer pages name files and symbols in backticks; user pages name what the application shows in bold and nothing from the code.
  • Say it once. A fact lives on one page; other pages link to it. Within a page, do not restate in a later section what an earlier one established.
  • Cut hedges and filler. "Note that", "it is important to", "basically", "simply", "you might want to" go. State the rule or the fact.
  • Caveats next to the feature, in a sentence, not in a separate Notes or Limitations section.
  • Tables for parallel structured items (settings, questions, files, triggers), prose for argument. A table cell holds a phrase, not a paragraph; if cells grow past two lines, the content wants prose or its own section.
  • Diagrams for relationships that prose hides. Use a small Mermaid flowchart, sequence diagram, state diagram, or class diagram when the reader must track at least three components, branches, transitions, or parallel and sequential phases. For dense architecture and lifecycle views that need deliberate layout, add a theme-aware SVG under docs/diagrams/ and embed it as <Diagram name="..." caption="..." />; follow the accessibility and styling rules in docs/developer-guide/docs-guidelines.md. Keep labels concrete, explain the consequence in nearby prose, and make the surrounding text sufficient for readers who cannot see the diagram. Do not diagram a list, a single linear action, or the same information already clear in a table.
  • Keep markdown compact. Tables use | --- | separators without column padding; docs are not run through a formatter in CI, so match the surrounding file.

Verify against the code

A documentation claim is a claim about the code. Before a page is done:

  1. Open every file and symbol the page names and confirm they exist with that name and role.
  2. Check every stated behaviour (default values, limits, orderings, fallbacks, error handling) against the implementation or its tests, not against memory or an older page.
  3. For a user page, confirm page and button labels against the UI components.
  4. Resolve every relative link; a renamed or moved page needs its old path in docs/.vitepress/theme/redirects.ts.
  5. A new page gets a sidebar entry in docs/.vitepress/config.mts, in reading order, and a one-line link from the page a reader would arrive from.
  6. Run bun run docs:build in docs/ when links or the sidebar changed; it fails on a dead internal link.

When a check fails, fix the page, not the code, unless the task is about the code.

Review

Read the page once as its reader, with these questions, and rewrite until every answer is yes:

  • Does the first screen tell the reader whether this is the page they need?
  • Can the reader find a specific answer (a setting, a file, a step) without reading from the top? Headings and the first sentence of each paragraph must make that possible.
  • Does each paragraph add one thing the page did not already say?
  • Is the order the order the reader needs: what it is, how it behaves, how to change it, where the code is?
  • Would a compact diagram make a multi-component relationship, branch, state change, or execution order faster to understand than the current prose? If so, add one and remove duplicated prose.
  • Would every example run, and does every path, symbol and label exist?
  • Can any sentence be deleted without losing a fact or a reason? Delete it.
  • Is the page still under about 400 lines? Past that, split by topic and link.

Report what was verified against the code and what could not be verified.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.