00 AI basics for developers
Tokens, LLMs, context window, harness engineering
12 sub-modules Free Passed
Claude Code Course
6 months ahead, in a few hours.
Ship in a day what your teammates ship in a week. It's not magic, it's a system. It's called harness engineering.
17
modules
+200
code examples
Free
to get started
$ claude "Implement ticket PROJ-247"
↳ MCP jira.get_issue PROJ-247
PROJ-247 Refactor auth module
Priority: High · Assignee: thomas.p · Sprint 34
↳ Skill activated: feature-dev
reading src/lib/auth.ts (142 lines)
reading tests/auth.spec.ts (88 lines)
src/lib/auth.ts
- export function verifyToken(token: string) {
+ export async function verifyToken(
+ token: string,
+ options?: AuthOptions
+ ) {
↳ Hook post-edit: lint + tests
✓ eslint 0 errors
✓ vitest 24 passed
↳ Commit feat(auth): typed verifyToken + options
3 files · +47 −12 · 1 commit
$
Opus 4.6 (1M context)
USED BY ENGINEERS AT
AlanAccorQontoSwileLydiaDecathlonBlaBlaCarSNCF
§ 02 · THE METHOD
FUNCTIONAL SCHEMATIC
Six layers, one loop.
The harness is the layer wrapped around the model: hooks, permissions, tools, memory. It is what separates an LLM that answers from a real collaborator. We show you, layer by layer, how to set it up on your codebase in one afternoon.
- INPUT
Intent
Translate a human need into a clear contract for the machine. No hacked prompts, briefs that hold up.
- Canonical brief
- Acceptance criteria
- Doubt protocol
- MEMORY
Context
Give Claude the same view you have: codebase, tickets, conventions, past decisions. A million tokens, well spent.
- CLAUDE.md
- Semantic index
- Decision log
- MCP
Tooling
Wire up the systems that matter: Jira, GitHub, staging database, Linear, Figma.
- MCP servers
- Scoped auth
- Audit trail
- SKILLS
Skills
Skills = activatable know-how. A style, a procedure, a coded checklist. You pick the mode, not the words.
- feature-dev
- deep-review
- post-mortem
- ops
- HOOKS
Guardrails
Every Claude keystroke, a hook verifies. Lint, tests, types, security policy. Error → Fix
- pre-tool
- post-edit
- pre-commit
- policy
- SHIP
Shipping
Signed PR, changelog written, ticket closed. The loop closes on itself.
- Draft PR
- Auto changelog
- Deploy preview
- HUMAN LATENCY
- 12 min to scope
- CLAUDE LATENCY
- 4 sec per hook
- COVERAGE
- 100% tests by default
- OUTPUT
- PR reviewed · merged
17 modules. +200 code examples.
Everything you need to master Claude Code.
From AI beginner to augmented developer. Each module unlocks a new superpower.
01 Models & CLI overview
Claude, GPT, Gemini, Codex CLI, Cursor, Copilot
13 sub-modules
Pro
Passed
02 Claude Code fundamentals
Installation, agentic loop, CLAUDE.md, plan mode
12 sub-modules
Pro
Passed
03 Advanced configuration
Settings, permissions, sandbox, memory, models
10 sub-modules
Pro
Passed
Do in one hour
what others do
in one day.
Skills, hooks, MCP, subagents — the full harness to industrialise Claude Code.
Course One-time payment
89€
Full access, updates included. For life.
- 17 complete modules
- +200 code examples
- Lifetime updates
- Instant access
Tailored
6 months ahead, for your entire team
Your team already uses Claude Code. Let's take it to production-grade.
- Audit of your practices
- On-site or remote training, tuned to your stack
- 1:1 coaching for your SWEs
- Skills, hooks and routines setup