One clear next step
You do not have to complete every language, install every runtime or use every agent feature. Choose a path, prove its prerequisites and move on when the evidence is ready. Course levels describe complexity, not a certification.
[!TIP] Recommended first route: Start here → lab 01 → one language through labs 02–05 → customization → one advanced scenario. If you prefer short fantasy stories, take the adventure route instead.
What is required?
| Label | Meaning | Example |
|---|---|---|
| Required | Needed for the chosen exercise’s acceptance criteria | Its runtime, baseline, scoped change, negative check, review and reset |
| Optional | Adds another experience without blocking the local lesson | A GitHub repository, second language, Codespace, UI or live integration explicitly labeled optional |
| Advanced | Requires earlier mental models and additional review | External tools, parallel ownership, SDK identity, cloud execution and modernization |
| Maintainer | Changes this curriculum, not the learner’s application | Site builds, packaging, translations and publishing |
Copilot access is required for live AI practice, not for inspecting source or running deterministic local checks. If access is unavailable, record that limit instead of presenting a simulation as a real agent run.
Route A — Professional hands-on practice
Stage 1: orient yourself
- Complete Start here for one verified change.
- Complete 01 — Context and interface.
- Explain role, harness, environment and permissions using your own session.
Exit evidence: a passing baseline, a reviewed change, a negative assertion and an honest record of which tools were available.
Stage 2: choose one application language
| Order | C# route | Python route | Why this comes next |
|---|---|---|---|
| Prepare | C# environment | Python environment | Select the correct runtime and working directory |
| 02 | Analyze/document | Analyze/document | Understand the existing application before changing it |
| 03 | Develop a feature | Develop a feature | Connect acceptance criteria to an end-to-end behavior |
| 04 | xUnit tests | pytest tests | Learn whether tests actually detect a wrong result |
| 05 | Safe refactoring | Safe refactoring | Change structure while preserving observable behavior |
Each kit starts from its own declared baseline. Your solution from the previous exercise is not silently required by the next kit. Do not overwrite one copy with another; preserve the evidence from each attempt.
Stage 3: practice an engineering decision
Choose the scenario that matches your work. They are not seven prerequisites for every advanced lab.
| Labs | Scenario | Preparation |
|---|---|---|
| 06 | Accessible shopping prototype | Node, browser, small PRD |
| 07, 08, 09 | Duplication, extraction and decision rules | C# basics and characterization tests |
| 10 | Bounded profiling | C#; a small workload and unchanged functional output |
| 11 | Issue to reviewed change | Local regression test; remote PR optional |
| 12 | Credential-free incident simulation | Node; no real token or provider operation |
Stage 4: customization and advanced scenarios
Start with 15 — Customization once you can verify a small change. Then select a purpose, not simply the next number:
| Purpose | Sequence | Evidence boundary |
|---|---|---|
| Specify a new product | Spec Kit setup → 13 — Greenfield | Local RSS contract; no remote feed fetching |
| Add behavior to an existing app | Setup → 14 — Brownfield | Old and new contracts pass together |
| Change storage without changing consumers | Setup → 17 — Modernization | Data reconciliation, non-overwrite and rollback |
| Embed an agent runtime | SDK setup → 16 — SDK | Offline tool/lifecycle tests; authenticated inference recorded separately |
Route B — Story-supported adventures
Use the adventure map in this order:
| Stage | Sequence | Ready to continue when… |
|---|---|---|
| Foundations | Portals of Nexus → Context Mirrors | You separate role, harness and environment and can ground an answer in files |
| Basics | Tempora Loop → Laws of Eldoria | You can bound iteration and write scoped, testable instructions |
| Intermediate | Skills of Algora → Agents of Stellaris → Guardrails of Stonevale | You distinguish expertise, role, handoff and deterministic policy |
| Advanced | MCP Cartographer → Lumoria Graph → Parallel Mythos | You can describe trust boundaries, dependencies and independent ownership |
| Surfaces | Cloud Citadel → Terminal Gate → Automaton Foundry | You distinguish contract validation from actual remote/runtime execution |
| Capstone | Convergence of Three Realms | You can map requirements to actors, environments, artifacts and reviewed evidence |
Every adventure includes a corresponding lab, a failure exercise and a rubric. Download one kit at a time. The story is a memory aid; correctness still comes from the lab’s acceptance criteria.
What “basic to advanced” does and does not mean
| Covered by this kit | Not automatically established |
|---|---|
| Investigation, planning, implementation, testing and review | Mastery of every programming language or every Copilot feature |
| Local exercises for instructions, skills, agents, MCP and parallel tasks | Universal host support or permission to install arbitrary servers |
| Spec-driven greenfield, feature and modernization workflows | A production migration or support for every data platform |
| SDK boundaries and optional live requests | Model quality across real users or production-scale traffic |
| Cloud-task contracts and optional authorized delegation | A cloud run merely because a JSON verifier passed |
| Accessible learning graphics and UI checks where relevant | A blanket accessibility certification of every student solution |
The kit teaches a method and supplies evidence-producing exercises. Additional production concerns—operations, compliance, real identity, deployment and broader evaluation—require their own requirements and verification.
Pick a realistic stopping point
- First session: finish the greeting exercise and explain your evidence.
- Application practice: complete one language’s 02–05 sequence.
- Team workshop: choose one engineering scenario, then a reviewed handoff.
- Advanced practice: complete one integration scenario and label local versus live evidence.
Do not rush to the capstone by skipping a failed prerequisite. Use the setup guide for environment blockers and the lesson’s troubleshooting for behavior failures.
| Previous | Next |
|---|---|
| Start here | Download your selected kit |