skills/understanding-first/SKILL.md
When generating wiki pages, READMEs, POC docs, integration write-ups, architecture notes, design docs, or any artefact whose purpose is for the user to *understand* a topic — not just record information. Apply BEFORE writing the artefact, then validate against the checklist before saving. The "outsource thinking, not understanding" principle.
npx skillsauth add RonanCodes/ronan-skills understanding-firstInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Outsource thinking. Don't outsource understanding.
Generated docs must teach the reader, not just record facts. When a reader (you, a future agent, a teammate) is done with the page, they should know why the thing exists, how it works at a glance, and where to dig deeper. If they only learn what the page is about, the doc failed.
Auto-load when generating:
Don't apply to:
coding-principles instead)Every doc covered by this skill must include the seven elements below, in this order where practical.
| # | Element | Why | |---|---------|-----| | 1 | Plain-English explainer (lead paragraph) | The "what is this and why does it exist" before any detail. Reader should be able to stop here and have the gist. | | 2 | Acronym handling (inline-on-first-use + glossary if dense) | Every acronym expanded the first time it appears. No "of course you know what an ORM is." | | 3 | Mermaid diagram of flow or structure | Diagrams compress 200 words of prose. One per page minimum for any flow or relationship. | | 4 | Concrete worked example | One real scenario walked end-to-end with real values, not placeholders. Examples teach faster than abstractions. | | 5 | Citations (sources of facts) | Every non-obvious claim cites where the fact came from: vendor docs URL + version, API spec endpoint, source file path, meeting date. | | 6 | Further reading (links to learn more) | External links to vendor docs, blog posts, talks, related concept pages. Distinct from citations: citations are where this fact came from; further reading is where to go to deepen understanding. | | 7 | Open questions (explicit section) | Anything not known goes in its own "Open" section with the question + who can answer it + how to chase it. Never bury uncertainty in prose. |
First use of any acronym, format:
SSR (Server-Side Rendering — pages rendered on the server, not in the browser)
RSC (React Server Components — server-rendered component model used in Next.js / TanStack Start)
ORM (Object-Relational Mapper — e.g. Prisma, Drizzle)
ADR (Architecture Decision Record)
MOC (Map of Content — an Obsidian convention for hub pages)
For docs with 4+ distinct acronyms, add a Glossary section at the top, ordered alphabetically, one line per acronym.
| | Citations | Further reading |
|---|---|---|
| Purpose | Where this fact came from | Where the reader can deepen understanding |
| Format | Inline links + sources: frontmatter | Footer section ## Further reading |
| Required? | Yes, for every non-obvious claim | Yes, ≥3 links for any tech topic |
| Examples | Vendor docs URL + version, OpenAPI spec POST /webhooks v2.4, screenshot path raw/<source-slug>/screen.png, dated meeting note | Vendor developer docs, related blog post, conference talk recording, vendor's GitHub repo |
For any third-party technology mentioned (e.g. your auth provider, database, payments processor, file-storage vendor, message queue) include at least one online further-reading link so the reader can verify or extend.
A "gold-standard" understanding-first doc lets a stranger become productive in one sitting. They leave knowing why the system exists, how the major pieces fit, and where to go next — without follow-up questions. Karpathy's tutorial-style READMEs (e.g. nanoGPT) are a public example of this pattern. Use any similarly-written OSS doc as a model.
A page titled "Webhooks" containing:
We use a payments provider for checkout. There are 2 webhook events
we care about. The signing secret is stored somewhere. See the wiki.
What's wrong:
Before saving any doc covered by this skill, verify:
sources: frontmatter)If any box is unticked, the doc is not understanding-first. Revise before saving.
Default to a consistent palette across a project for visual consistency. A balanced default:
#e0af40 for user / source nodes#5bbcd6 for engine / skill / system nodes#7dcea0 for output / result nodesAdjust per project's brand palette if one exists.
Pages produced under this contract should carry frontmatter that makes citations and further reading machine-readable:
---
title: <human-readable title>
page-type: integration | concept | architecture | adr | onboarding | poc-explainer
sources:
- "docs-cms:<space>:<page-id>@<version>"
- "openapi:<api-name>:<endpoint>"
- "raw:<source-relative-path>"
- "meeting:<date>:<topic>"
further-reading:
- url: https://...
title: "..."
why: "Vendor docs for the API used in this integration"
open-questions:
- question: "Which endpoint owns this data — A or B?"
owner: "<person to chase>"
raised: <YYYY-MM-DD>
---
Skills that surface "everything with open questions" or "everything citing docs-cms page X" rely on this shape — keep it consistent.
testing
--- name: linear-pipeline description: The Fable orchestrator for a single dispatched Linear ticket. Holds almost no context itself; it receives `--issue <ID> --detached`, decides the stage sequence, and fans out a sub-agent per stage, passing forward only each stage's artifact (never re-derived, never inlined into its own context). Step zero, before any planning or stage routing, is a boundary triage against `canon/security-boundary.md` (#199): a match tags Ronan Connolly and stops the run, no
development
--- name: in-your-face description: Capture a chat-only answer into a durable artifact (markdown + HTML, PDF when cheap) and launch it automatically so the user cannot miss it. Use when user says "in your face", "don't let me lose this", "save that answer", "make that durable", or right after answering a substantive side question (a recipe, comparison, how-to, or generated prompt) that would otherwise die with the context. category: workflow argument-hint: [--no-open] [--vault <short>] [hint of
tools
One-shot headless OpenAI Codex CLI calls for background/admin AI tasks — summaries, classification, extraction, admin glue. The default engine for anything that runs AI constantly in the background (daemon-driven, per-event), because it bills the flat ChatGPT subscription instead of Claude usage or per-token API spend, and it keeps working while Claude is rate-limited. NEVER for coding — coding stays Claude. Use when a skill or daemon needs a cheap always-on AI call, when the user says "use codex", "ask codex", "codex as backup", or when building a background summarizer/classifier into a listener or loop. Reads auth from ~/.codex/auth.json (ChatGPT account, no API key).
research
Turn a warranty rejection, repair quote, or RMA email into a cited decision brief — legal read (NL/EU consumer law), is the part user-serviceable, live part and new-unit prices, repair-vs-DIY-vs-new economics, before-you-send-it checklist, deadlines. Use when the user pastes or screenshots a repair quote, warranty rejection, "not covered" email, onderzoekskosten fee, or asks "should I repair or replace this".