Workstation Setup tutorial series — Part 6: Done — what now?
Your workstation is complete. This is the lap of honour — a short map of the next tutorials, grouped by what kind of work you'll do. OpenSpec, Claude Skills, Build an App, Hydra — what each one is for, and where to start. Sixth and final of six short modules.
Your workstation is complete. WSL2, Docker, VS Code with the right extensions, Claude Code with the safety hooks live, the MCP browser pool connected, Nextcloud running locally with the apps you need. Take a moment to feel good about that — most new Conduction developers used to lose one or two days to this; you've done it in one evening. This last part is a short map of the next tutorials, grouped by what kind of work you'll do.
The four pillars
Almost every Conduction tutorial leans on one of four pillars. Pick the one that matches your first real task:
Which one first?
Honest answer: OpenSpec, then either Skills or Build an App, then Hydra last. The order matters because:
- OpenSpec is the substrate. Every change to every app — including the next four tutorials' worth of changes — is shaped as an OpenSpec change. Skipping it makes the rest harder to read.
- Skills and Build an App are sideways from each other. Pick the one closer to your first task.
- Hydra is the most opinionated and the most internal. Save it for when you're confident with the previous three.
If your first task involves writing actual code
Start with OpenSpec Part 1 (~12 min) — what OpenSpec is, the four core concepts. Then OpenSpec Part 2 (~22 min) — your first change end-to-end.
After that, jump to Build an App tutorial — Part 0 for the honest map of how Conduction apps actually get built (three paths, why we teach the manifest path). Then walk Parts 1–8 — scaffold, schemas, calendar, integrate, ship — using the PetStore as the running domain.
If your first task involves writing tooling or skills
Start the same way — OpenSpec Part 1 and Part 2 — because skills are versioned and specified the same way as app code. Then go to Claude Skills Part 1 (~12 min) — what a skill is, when not to write one. Parts 2–4 of that series take you through writing your first skill and the maturity framework that turns "feels good" into "measured good".
If your first task is about Hydra itself
You still want OpenSpec under your belt first. After that, Hydra Part 1 (~10 min) — what Hydra is, four container personas, when not to use it. The series has six parts; budget an afternoon for the full read.
What's not covered by these tutorials
A few areas you'll meet at Conduction that don't have a dedicated tutorial series yet:
- The day-to-day Claude Code workflow — the getting-started.md doc in
Conduction/.githubon Codeberg is the closest to a single-page workflow primer. - Frontend standards — covered in frontend-standards.md. Worth a skim before your first Vue PR.
- Testing skills in depth —
/test-app,/test-counseland the rest of the testing workflow are documented in testing.md. Thehydra-gate-*family lives inhydra/.claude/skills/in the hydra repo — closer to the gates themselves. Read on demand, not upfront. - The local LLM workflow (Ollama + Qwen) — see local-llm.md. Optional, useful if you want overnight batch jobs.
These are reference docs, not tutorials. Use them when a specific question comes up; don't try to read them upfront.
A note on the quiz blocks
Every part of this series ends with a Test yourself quiz block — a few short questions with Hint and Answer fold-outs. They're not exam questions; they're a self-check that the part actually landed. If you blank on most of them, the part is probably worth a re-read.
For this series specifically, the questions cover:
| Part | Sample question |
|---|---|
| 1 — Overview | Which OS + tooling combination does Conduction recommend, and why? |
| 2 — Basics | Why install Node and PHP inside WSL rather than on the Windows side? |
| 3 — Claude Code | What's the difference between global and project-level Claude settings? |
| 4 — MCP | Which two settings pre-approve the MCP browsers so background agents aren't blocked? |
| 5 — Nextcloud | When to reset the local Nextcloud, and when to keep the existing state? |
| 6 — What now? | (the questions below) |
Spend a few minutes on the quiz blocks before declaring a part "done" — that's the easiest way to make this knowledge stick.
How long does this all take?
A rough estimate, based on the reading times the tutorials themselves list:
- This workstation series: one evening (the six parts together are well under an hour of reading, but plan for installs and the first run of each piece).
- OpenSpec Parts 1+2: ~35 minutes of reading (12 + 22), plus the time to actually do the first change.
- Claude Skills Parts 1+2: ~30 minutes of reading (12 + 18). The full four-part track adds Parts 3 and 4 on top.
- Build an App tutorial Part 0: ~15 minutes for the architectural overview. The full 9-part series spans multiple evenings.
Add a Conduction colleague pairing with you on a first real change, and "clean laptop to first merged PR" is realistically a few evenings of mixed reading and doing. Your mileage will vary with how much terminal and Git muscle memory you bring in.
Test yourself
Three short questions to check that this part landed. Stuck? Click Hint. Curious about the answer? Click Answer.
1. What's the first tutorial series you do after finishing this Workstation Setup chain — and why that one?
Hint
It's the substrate underneath everything else.
Answer
The OpenSpec tutorial series — two short parts, ~30 minutes total.
Why first: every change to every Conduction app — including the tutorials after OpenSpec — is shaped as an OpenSpec change. Skipping it makes the rest of the academy harder to read because you'll be guessing at the structure of every proposal.md, tasks.md and spec-delta you see.
2. Your first task is to write a Claude Skill. You're tempted to skip OpenSpec and dive straight into the Skills series — after all, the task isn't about app code. Why is that probably a mistake?
Hint
How does a skill itself get tracked, reviewed, and changed at Conduction? Look at how the existing skills in hydra/.claude/skills/ are version-controlled.
Answer
Because skills are versioned and specified the same way as app code. Every skill at Conduction lives in .claude/skills/<name>/SKILL.md, and a non-trivial change to a skill goes through the same OpenSpec workflow as an app feature: proposal → spec-delta → tasks → review.
If you skip OpenSpec, the first time you propose a skill change you'll see the reviewer ask "where's the spec-delta?" and you'll be back-filling concepts that 30 minutes of OpenSpec Parts 1+2 would have given you upfront. The Skills tutorial series even references proposal.md, tasks.md and spec-delta as if you already know what they mean — because it assumes you've done OpenSpec first.
Spec-first applies to everything Conduction ships, including the tooling that ships the things. Skipping OpenSpec for skill work is like skipping Git for solo work — you can do it, but you're paying the cost later in a less convenient form.
3. Why is the Hydra tutorial series saved for after OpenSpec, Skills, and Build an App?
Hint
Two reasons — one about prerequisites, one about scope.
Answer
Two reasons:
- Hydra builds on top of OpenSpec and Skills. It runs
/opsx-*skills against OpenSpec changes inside containers. Without those two pillars first, the Hydra tutorials read like a foreign language. - It's the most internal and most opinionated of the four pillars. Many new developers won't touch Hydra in their first month; the ones who do should already be comfortable with the rest of the stack.
So: OpenSpec → Skills (or Build an App) → Hydra. If you want a one-pager preview before reading the full series, the hydra/README.md in the .github repo is a good warm-up.
Next steps
You're done with the workstation series. Below are the four pillars in the order most new developers actually do them.