Awesome CopilotAdventures
Language
Start learning

Agentic engineering · Evidence first

Learn the craft.
Prove the result.

A complete learning library for GitHub Copilot: guided adventures, practical hands-on labs and the original source behind every exercise.

Start learning Download a lab

Ask → Plan → Agent → Review → Evidence

14Adventures26Hands-on labs41Guides1561source files

Learning navigation

New here? Follow one clear path.

Start with one local check. Choose your next lesson, then download only the materials it needs.

Choose your format

Two ways in. The same care for evidence.

Follow a story or work through a professional lab. Both paths help you investigate, build and verify; you do not need to finish both.

Adventures

Fantasy supports the lesson. Learn context, custom agents, skills, hooks, MCP, loops, graphs, CLI, SDK and a capstone.

Explore this path

Hands-on labs

Numbered professional exercises. Practice software engineering, greenfield and brownfield Spec Kit workflows, and modernization.

Explore this path

Mechanism

A fluent answer is not a finished change.

Scroll sideways to see the whole diagram. With the diagram focused, use the left and right arrow keys.

Investigation leads to a plan, the plan bounds the implementation, a separate review challenges the result and evidence proves it. An unresolved gap returns to investigation instead of closing the work.ONE PROGRESSIVE WORKFLOW
Ask
investigate
Plan
design
Agent
implement
Review
challenge
Evidence
prove
UNRESOLVED GAP RETURNS TO INVESTIGATION
Bound the change first

A plan with acceptance criteria is what makes a result checkable later.

Review is separate work

The step that made the change is not the step that decides whether it is right.

Evidence closes the loop

Commands, output and diffs someone else can rerun, not a claim that it works.

The adventure path

Fourteen adventures, six levels, one progression.

Each level assumes the evidence from the level before it. Follow them in order, or jump to the capability you need.

  1. 01

    Foundations

    2 Adventures

    The Portals of NexusThe Context Mirrors
  2. 02

    Basics

    2 Adventures

    The Tempora LoopThe Laws of Eldoria
  3. 03

    Intermediate

    3 Adventures

    The Skills of AlgoraThe Agents of StellarisThe Guardrails of Stonevale
  4. 04

    Advanced

    3 Adventures

    The MCP CartographerThe Knowledge Graph of LumoriaThe Parallel Trials of Mythos
  5. 05

    Surfaces

    3 Adventures

    The Cloud CitadelThe Terminal GateThe Automaton Foundry
  6. 06

    Capstone

    1 adventure

    The Convergence of Three Realms

Generated visual stories

Adventures in motion

Three short montages introduce the adventure world. They are illustrations, not recordings of Copilot or evidence that a lab ran.

Hover or focus a scene to bring it to life. On touch screens, tap the image. Reduced motion keeps every scene still until you choose one.

0110 seconds · no audio

Portals, context and iteration

A navigator explores the Nexus workstations, compares the same mechanism in two context mirrors, then works around the circular mechanism of Tempora.

Open silent videoOriginal source

0210 seconds · no audio

From investigation to guardrails

Scenes move from a navigation desk to a clockwork loop, an archive city, a handoff between specialists and a controlled checkpoint with a stop mechanism.

Open silent videoOriginal source

0310 seconds · no audio

Rules, skills and connections

An archive charter gives way to a selected toolkit, specialist workstations, contrasting context mirrors and a cartographer observing a distant archive.

Open silent videoOriginal source

Interactive learning / local only

The practice studio

Make a decision. Observe the consequence. Reset and try again before taking the idea into a real lab.

01

Agent workflow

Walk through a bounded change and see why a failed review must stop the handoff.

Practice this concept
02

Context workbench

Choose just enough relevant context. More files are not automatically better context.

Practice this concept
03

Verification loop

Run a tiny JavaScript fixture in this browser. Compare a failing implementation with a corrected one.

Practice this concept

Educational simulations, not a live Copilot session. No AI model, account, terminal, file edit or external service is used.

Guides

The complete learning library

Every current learning document is available in English, Spanish and Brazilian Portuguese. Switch language without losing your place.

Open the complete library
The curriculum at a glanceGoals, prerequisites and the full getting-started reference.

Awesome Copilot Adventures

[!IMPORTANT] The curriculum teaches agent roles, harnesses, environments, and customization primitives as separate concepts. It does not teach deprecated custom chat modes.

Ask, Plan, and Agent are roles. Local and Copilot are VS Code agent harnesses. Cloud is a remote session target. A folder, worktree, Codespace, local machine, or remote ephemeral workspace is an environment.

[!NOTE] Product guidance was last verified against current official GitHub and Microsoft documentation on 2026-09-05. Recheck the linked source when availability or policy materially affects your task.

Search

Search in English. Source paths and executable examples retain their original text.