Master Claude·The Practitioner Curriculum

Master Claude Code.
From first prompt to autonomous agents.

The hands-on curriculum that turns developers into AI-native shippers. 133+ tutorials across 4 levels — Foundations through Expert — built by practitioners using Claude Code daily.

8
Courses
133+
Tutorials
5,900+
Minutes
Live agent pipeline
133 tutorials·8 courses·Updated weekly
01Prompt
Frame the task with rich context
02Plan
Break work into verifiable steps
03Code
Edit, test, iterate with the agent
04Review
Verify, refine, ship with confidence
05Ship
Deploy, observe, learn
$ claude "ship the new pricing page with tests"Stage 01 / 05
CLAUDE.mdSubagentsMCP ServersPrompt EngineeringTestingRefactoringAI TeamsSpec-Driven DevAutonomous LoopsToken OptimizationDebuggingCI/CD PipelinesBMAD MethodPlugin BuildingCLAUDE.mdSubagentsMCP ServersPrompt EngineeringTestingRefactoringAI TeamsSpec-Driven DevAutonomous LoopsToken OptimizationDebuggingCI/CD PipelinesBMAD MethodPlugin Building
Showcase

What you’ll actually build

Live animated scenes from across the curriculum. Auto-playing, hover-pausable, every beat backed by a tutorial.

Fundamentals

Stop writing "fix the bug"

A prompt rewrite is the single biggest leverage point in your day. Watch what context does to the same task.

claude — bad prompt
$ claude " "
claude — good prompt
$ claude ""
Tokens

Cut your team’s token bill 85%

Six techniques, stacked on the same workload. From $4,800/mo to $720/mo — same agents, same outputs.

Monthly bill · 5-engineer teamStage 01 / 06
$4,800/ mo
$4,800
BaselineNo optimization — every prompt re-sends the full context
Techniques applied
  1. 1Baseline
  2. 2Prompt caching−45%
  3. 3Subagent offload−18%
  4. 4Steered compaction−9%
  5. 5Prompt thrift−7%
  6. 6Plan selection−6%
Stacked total
0% saved this stage$0/mo
SDD & BMAD

When a requirement fails, the spec catches it

Watch the BMAD framework trace a multi-currency failure back to a silent spec gap, patch the spec, re-run green.

01 · Analyze
02 · Architect
03 · Build
04 · Verify
05 · Done
claude — ~/payments-platform
$ claude "BMAD: implement payment-gateway.md"
planning
Spec coverage0%
0/ 47
payment-gateway.md
R-01Idempotent createInvoice·
R-12Refund window 30 days·
R-23Multi-currency support·
R-31Stripe webhook retries·
R-47Audit log every state change·
+ 42 more requirements tracked
Fundamentals · L5

Build your own tool servers

From a one-line prompt to a registered MCP server in one session — Claude scaffolds the tools, you ship them.

claude — ~/mcp-jira-server
$ claude "Build an MCP server that connects to Jira. Tools: get_sprint_tickets, create_ticket, update_status."
planning
Created src/tools/get-sprint-tickets.ts
Created src/tools/create-ticket.ts
Created src/tools/update-status.ts
Created src/server.ts
Created claude_desktop_config.json
mcp-jira-server/0/5
src/
tools/
get-sprint-tickets.ts
create-ticket.ts
update-status.ts
server.ts
claude_desktop_config.json
Fundamentals · L4

Orchestrate three agents in parallel

A coder, a reviewer, a tester running side-by-side under a supervisor. Watch them fan out, converge, and ship.

$ claude /orchestrate "feature: invoices endpoint"spawning
t = 0s
Supervisor
spawn agents
agent-coder
agent-reviewer
agent-tester
3 agents·parallel where safein progress…
AI Teams

AI teammates review your PR

Diff, reviewer comments, tester sign-off, merge — the full review loop automated by agents that know your conventions.

src/billing/invoices.ts+2 / -1
export async function createInvoice(input: NewInvoice) {
const customer = await stripe.customers.retrieve(input.customerId)
- if (!customer) throw new Error("no customer")
+ if (!customer || customer.deleted) throw new InvoiceError("INVALID_CUSTOMER")
const idem = createIdempotencyKey(input)
+ span.setAttribute("invoice.idem_key", idem)
return stripe.invoices.create({ customer: customer.id, ...input }, { idempotencyKey: idem })
}
reviewing diff
🤖agent-reviewersuggest
Nice catch on the deleted-customer guard. Suggest naming the error constant.
🤖agent-reviewerapprove
Span attribute looks good — matches our tracing conventions.
🧪agent-testerapprove
Added test: throws INVALID_CUSTOMER when stripe returns deleted=true.
👤andersonapprove
LGTM. Merging.
CCA Architect

Practice the architect exam

A 60-question practice exam graded by official domain weights. Track where you’re strong and where you need another tutorial.

60-Question Practice Exam
Grading by domain — official weights appliedgrading
Running score
0/100
Architecture
MCP & Tooling
Agentic Patterns
Governance & Safety
Operations & Observability
·
Score in progress
0% —
More scenarios

A few more moments worth seeing

CLAUDE.md, auto-maintained

Fundamentals

Drift detection · spec patching · zero stale rules.

$ claude /audit-claude-mdauditing
Stage 00 / 05
SCANIndexed 47 rules across 3 CLAUDE.md files
DRIFT2 rules contradicted recent commits
PATCHRewrote stale rule on test runner
VERIFYReplayed last 3 sessions — no regressions
COMMITCLAUDE.md updated · ready for the team

Sub-agent dispatch

Hidden Power

Fan out three sub-agents, converge into a single PR.

MAIN AGENTclaude /ship feature/auth-refresh
planning fan-out
DOC-WRITER
README.md · CHANGELOG.md
TEST-AUTHOR
4 test files · 31 assertions
SECURITY-SCAN
No vulns · deps OK
MERGEawaiting sub-agents…
3× faster than sequential — the parallelization pattern from Course 6.

Crash-proof sessions

Hidden Power

Terminal closed? --continue restores 14 turns of context.

claude — recovered
terminal closed unexpectedly · 02:14 ago
Curriculum

Your Learning Journey

Six courses, one path. From zero to building autonomous AI systems.

1
Course 01Beginner → Expert

Fundamentals

Install Claude Code, write your first prompt, ship a real feature.

24TutorialsBeginner → ExpertLevel
2
Course 02Intermediate

Mastery

Real-world projects, enterprise patterns, production delivery.

12TutorialsIntermediateLevel
3
Course 03Advanced

SDD & BMAD

Spec-driven development and the BMAD framework for deterministic AI delivery.

19TutorialsAdvancedLevel
4
Course 04Intermediate

Tokens

Master token spend — caching, quotas, session hygiene, plan selection.

8TutorialsIntermediateLevel
5
Course 05Advanced

AI Teams

Human–AI coordination and organizational patterns for AI-native teams.

9TutorialsAdvancedLevel
6
Course 06Advanced

Hidden Power

The non-obvious shortcuts that make you dramatically faster than peers.

20TutorialsAdvancedLevel
Compare

The Old Way vs AI-Native

A fundamentally different way to build software.

How you work
★ BestAfter this curriculum
Without it
Structured AI-native workflows
Ad-hoc prompting
Multi-agent orchestration
Single-chat only
Spec-driven delivery pipeline
Hope-driven dev
MCP server building
No custom tools
Team-wide AI adoption
Individual heroics
Autonomous AI agent systems
Manual everything

Two ways to level up

Start free with the first three Foundations tutorials. When you're ready, pick a course or unlock everything with the monthly subscription.

Single Course

Pick Any Course

$20USD
one-time · 1 year of access

Lock in a single course for a year. Perfect if you know exactly what you want to learn.

  • Full access to one course
  • All tutorials & exercises
  • 1 year from purchase date
  • Course updates during your year
  • Other courses
  • Future courses
Get Started
MOST POPULAR
All Courses · Monthly

Everything, Always

$45/ month
cancel anytime

Unlock every course we offer plus everything we add next. The full journey from first prompt to AI-native team workflows.

  • All current courses
  • All future courses included
  • All tutorials & exercises
  • Cancel anytime, keep access until period ends
  • 1,000+ minutes of content and growing
Start Subscribing
Break even after 3 courses

First three Foundations tutorials are free with any account. Refunds available per our terms.

Code Differently

Not another tool. A fundamentally different way to build software.

Without structured AI training
  • Copy-pasting ChatGPT output
  • Fighting hallucinated code
  • No systematic approach
  • AI as autocomplete++
  • Slower than coding yourself
After this curriculum
  • Structured AI-native workflows
  • Autonomous agent orchestration
  • Spec-driven delivery pipeline
  • AI as your engineering team
  • 10x faster, higher quality

Frequently Asked Questions

No. Course 1 starts from zero — installing Claude Code, writing your first prompt. If you can write basic code, you're ready.

Tutorials average 30–60 minutes; most courses are ~10 hours of guided content (1–2 weeks at a steady pace). The Claude Certified Architect course is also ~10 hours of guided content but recommends ~84 hours of practice over 12 weeks to pass the exam.

Lifetime access to all tutorials, code examples, exercises, and updates for that course. 30-day money-back guarantee.

Start with Course 1 (Fundamentals). Then Course 2 or 3 depending on your goals. The timeline on the landing page shows the full recommended path.

Yes, you'll need a Claude Code subscription from Anthropic to follow along hands-on. The tutorials teach you to use Claude Code, so access to it is essential.

Absolutely. Course 4 is designed for team leads. Each member needs their own access. Contact us for volume discounts on 10+ seats.

The tutorials use Next.js and FastAPI as examples, but the skills apply to any language or framework. Claude Code works with any codebase.

Docs tell you what buttons exist. This teaches you how to think differently about building software — workflows, agent orchestration, and patterns no docs cover.

Popular tutorials & comparisons

Start with what most developers read first, or pick the comparison that matches your stack.

Stop coding
the old way.
Start shipping.

Join developers who transformed how they build software.

Get Started — From $20
133+
Tutorials
8
Courses
1,200+
Minutes
$20
Per Course
Try me!