Student guide
Prepare for arc42 in a Day
Everything you need before, during and after the day. All the reading below is free and open, and none of it is required: an hour of preparation is plenty.
01Before the day
Setup checklist
Your ticks are saved in this browser only.
02Preparation
Prepare at your level
New to architecture documentation
About 30 to 45 minutes:
- Skim the twelve arc42 sections and read what each one is for
- Try one Mermaid diagram in the live editor
- Think of a system you know well and who uses it
Experienced engineer or architect
About 60 to 90 minutes:
- Read the arc42 guidance for sections 5 and 9, and a sample ADR
- Bring a real constraint, quality goal and trade-off from your own work
- Look at one docs-as-code example and decide what you would change
03Reading
Free and open material
- arc42 documentationThe official guide to each of the twelve sections, with tips.
- arc42 FAQShort answers to common questions, including when not to use arc42.
- Why arc42 and how it is structuredAn overview of the template and where everything lives.
- The arc42 template on GitHubGet it in Markdown or AsciiDoc and keep it in your repository.
- arc42 examplesWorked examples of real documentation.
- arc42 quality modelQuality labels and example scenarios to start from.
- The C4 modelContext, containers, components and code: the zoom levels we use.
- MADRA lean Markdown template for architecture decision records.
- Mermaid documentationText-based diagrams: graphs and sequence diagrams.
- Mermaid live editorTry diagrams in your browser with nothing to install.
- Markdown basic syntaxHeadings, lists, tables and code blocks in one page.
- Architecture as Code, an exampleOne team’s approach combining arc42, C4, ADRs and Mermaid.
- AGENTS.mdAn open format for giving coding agents project instructions.
- What goes in AI context filesA practical write-up on keeping them short, citing a 2026 ETH Zurich study.
Links point to third-party sites that we don’t control. If one has moved, searching for its title will find it.
04Outputs
What you will produce
| Exercise | Produces | Section | Done when |
|---|---|---|---|
| 1 | Summary, quality goals, stakeholders | 1 | Three goals, each with a numbered scenario |
| 2 | Constraints | 2 | Each passes the three-question test |
| 3 | Context view | 3 | Every arrow is labelled, no internal parts |
| 4 | Strategy table | 4 | A neighbour can guess your architecture |
| 5 | Level-1 blocks and one contract | 5 | 5 to 9 blocks and at least one error mode |
| 6 | One runtime scenario | 6 | Includes a failure branch |
| 7 | Deployment sketch | 7 | Mapping table is complete |
| 8 | Crosscutting concepts | 8 | Each concept spans two or more blocks |
| 9 | One ADR | 9 | Two options and one negative consequence |
| 10–11 | Quality scenarios, risks, glossary | 10–12 | Numbers, mitigations and owners |
| 12 | AI-drafted section, red-teamed | Any | Every AI line has a verdict |
05Template
Team skeleton
Paste this into arc42-team.md on the day.
# <System name> — arc42 skeleton
## 1. Introduction and Goals
### 1.1 Requirements overview
### 1.2 Quality goals (top 3–5, each with a scenario)
### 1.3 Stakeholders (role, expectation)
## 2. Constraints (ID, constraint, why, effect)
## 3. Context and Scope
### 3.1 Business context (diagram)
### 3.2 Technical context (table)
## 4. Solution Strategy (goal, approach, why, trade-off)
## 5. Building Block View
### 5.1 Level 1 (diagram + table)
### 5.2 Contract: <block>
## 6. Runtime View (scenario + failure branch)
## 7. Deployment View (environments, mapping table)
## 8. Crosscutting Concepts
## 9. Architecture Decisions (ADR-0001…)
## 10. Quality Requirements (tree + scenarios)
## 11. Risks and Technical Debt
## 12. GlossaryMermaid in 30 seconds
```mermaid
graph LR
A[Customer] --> B[Order app]
B -->|HTTPS| C[Payment provider]
```
```mermaid
sequenceDiagram
Customer->>App: place order
App-->>Customer: confirmation
```
06Vocabulary
Key terms
- Stakeholder and concern
- A stakeholder is anyone with an interest in the system. A concern is something they care about, such as recovering a lost order.
- Quality scenario
- A testable statement of stimulus, response and metric, such as “95% of confirmations appear within 3 s at 300 orders per minute”.
- Container and component
- A container is something you start: an application or data store. A component is a grouping of related functionality behind an interface inside a container.
- Contract
- What a block promises and expects: pre-conditions, post-conditions, invariants and error modes.
- ADR
- An architecture decision record: one significant decision with its context, options, decision and consequences.
- Idempotency
- Repeating a request has the same effect as doing it once.
07Ground rules
Rules for AI today
Write a human draft first. Then ask AI why, and what’s missing, not only to write it. You decide: give every AI line a verdict of accepted, rejected or edited, with a reason. Use only your workshop case, and never paste passwords, keys, personal data or client-confidential details.
08Self-check
Check before you leave
09After
Your 30 days
Week 1: finish sections 1–4 with your real team. Week 2: sections 5–6, one contract per critical block. Week 3: three ADRs, five scenarios, risks. Week 4: try the agent experiment (no docs, a short index plus two contracts, or the full arc42) and share what you find.
Not registered yet?
Saturday 30 January 2027, UWI St Augustine.