Claude Code Workshop — Canonical Curriculum v1.5

⚡

v1.5 change: Week 2 and Week 3 swapped — participants build first (Week 2), then customise once they've felt the friction (Week 3). Hooks demoted from graduation requirement to optional/bonus. Feedback: Raffaello, 2026-06-20.

WEEK 1Understanding AI + Claude Code Essentials

📖Before classASYNC

~55 min

#	Subtopic	Type	Time	What to do
1	How AI works	Reading	30 min	Probabilistic AI · context is not storage · AI tool landscape · strengths and limits
2	Why Claude Code is different — the harness	Reading	10 min	The harness: read, write, run, check, loop. Read before class so the demo lands as recognition.
3	Quiz — answer before class	Quiz	10 min	5 questions on both readings. Answer in your own words, check model answers.
4	Instructions — how to prepare	Instructions	5 min	Install Claude Code CLI. Have a real work folder open during class.

🎥In classLIVE

57 min + ~15 min FAQ

#	Subtopic	Type	Time	What happens
1	Recap of reading	—	5 min	"You read about the harness — today you watch it run."
2	The agentic loop	Demo	15 min	Instructor runs Claude Code live: two sequences, a file created. First time participants feel it.
3	The approval gate	Demo	12 min	Instructor shows plan → reads it → redirects it. Participants classify and redirect a second plan.
4	System prompts / CLAUDE.md concept	Concept	5 min	Why persistent instructions matter. Primes the after-class exercise.
5	Tier classification	Demo	8 min	Instructor classifies 5 real tasks with reasoning. Participants do the exercise async.
6	Validate before fan-out	Demo	7 min	Multi-file plan without canary → disaster. With canary → caught at step 1.
7	Session discipline	Demo	5 min	Instructor creates todo.md and saves state live. "This is what you do after every session."
—	FAQ	—	~15 min	—

✏️After classASYNC

~40 min · share before Week 2

#	Exercise	Type	Time	What to do
1	Write your CLAUDE.md	Exercise	25 min	Open real work folder. Write WHY / Constraints / HOW. Under 60 lines. Share before Week 2.
2	Tier classification	Exercise	15 min	10 real tasks from your work. Classify T1/T2/T3 with one sentence reasoning. Share before Week 2.

WEEK 2Build a Feature

Build first. The "aha — I shipped something" moment converts fragile beginners into committed operators. Skills and rules come next week, when the friction is real.

📖Before classASYNC

~55 min

#	Subtopic	Type	Time	What to do
1	Define done before you start	Reading	10 min	The 3-criteria rule: before any code, define 3 verifiable success criteria. "It looks right" is not a criterion.
2	Design preview before code	Reading	10 min	Sketch in plain text before asking Claude to list files. Catches misunderstandings before a diff exists.
3	Consumer sweep	Reading	10 min	When you change a convention, grep every dependent file before calling done. Highest-recurrence error in the map.
4	Quiz — answer before class	Quiz	10 min	5 questions on the readings.
5	Instructions — how to prepare	Instructions	15 min	Bring your CLAUDE.md from last week. Pick one real feature to build (small: one button, form, or script).

🎥In classLIVE

56 min + ~19 min FAQ

#	Subtopic	Type	Time	What happens
1	Recap of reading	—	5 min	"You read the 3-criteria rule — today you write criteria live before we build anything."
2	Show and tell — Week 1 CLAUDE.mds	Demo	5 min	Instructor picks 2-3 participant CLAUDE.mds. Shows what strong looks like.
3	Define done — gate	Exercise	8 min	Participants write 3 success criteria for their feature live. Instructor doesn't proceed until everyone has 3. Hard gate.
4	Design preview	Exercise	10 min	Participants sketch their feature in plain text. Instructor reviews 2-3 sketches with the group.
5	Build-feature walkthrough	Demo	20 min	Instructor runs a complete feature build: preview → scan → branch → build → self-review → deploy.
6	Compounding loop ritual	Demo	8 min	Instructor writes one capture live. "What went wrong? One sentence." Shows format and location.
—	FAQ	—	~19 min	—

✏️After classASYNC

~65 min · share before Week 3

#	Exercise	Type	Time	What to do
1	Build and ship your feature	Exercise	45 min	Use criteria and sketch from class. Canary check. Consumer sweep. Deploy and verify. Share PR link + criteria marked ✓.
2	First compounding entry	Exercise	20 min	One real mistake-log entry: what happened / why / what you'll do differently. Share before Week 3.

WEEK 3Skills, Rules & MCPs

Customise once friction is real. Participants have now built something manually — the 3-rep heuristic is meaningful, the MCP friction-twice gate is testable against real experience.

📖Before classASYNC

~50 min

#	Subtopic	Type	Time	What to do
1	Rules files concept	Reading	15 min	What rules files are, why they're separate from CLAUDE.md, paths: vs globs: — and why the wrong one silently breaks 20 files.
2	Skills concept	Reading	15 min	The 3-rep heuristic. Context-first pattern. Only write a skill for a workflow you've already done manually — you have some now.
3	MCPs concept	Reading	5 min	The friction-twice gate. You've been building for two weeks — have you hit the friction of not having a specific tool?
4	Quiz — answer before class	Quiz	10 min	5 questions on the readings.
5	Instructions — how to prepare	Instructions	5 min	Bring your shipped feature and compounding entry. Think of one workflow you've repeated manually that could become a skill.

🎥In classLIVE

47 min + ~28 min FAQ

#	Subtopic	Type	Time	What happens
1	Recap of reading	—	5 min	"You read the 3-rep heuristic — today we build a skill for a workflow someone in this room already did manually in Weeks 1-2."
2	Self-audit: CLAUDE.md review	Demo	5 min	Instructor opens a fresh Claude session and reviews one participant's CLAUDE.md. Group sees what fresh context finds.
3	Rules files	Demo	10 min	Wrong config (globs:) fails silently — 20 files, zero warnings. Then correct config (paths:).
4	Skills	Demo	12 min	Instructor builds a skill for a workflow a participant described — sourced from their actual Weeks 1-2 work.
5	MCPs: decision gate	Exercise	8 min	Group exercise: each participant names a tool they've hit friction with twice in the last two weeks. Real decisions now.
6	Hooks — optional/bonus optional	Demo	7 min	Brief demo: rule ignored vs hook catches it. Positioned as "infrastructure for when you're ready." Not in graduation requirement.
—	FAQ	—	~28 min	—

✏️After classASYNC

~65 min · share before Week 4

#	Exercise	Type	Time	What to do
1	Write your first rule	Exercise	20 min	Write one constraint rule using paths: (not globs:). Verify it auto-loads. Share content + confirmation.
2	Write your first skill	Exercise	30 min	Write a skill for a workflow you already did manually in Weeks 1-2. Real work only. Share skill file + screenshot running.
3	MCP setup optional	Setup	15 min	Only if you identified a friction-twice use case. Install one MCP. Share what it connects and why.

WEEK 4Agentic Skills

📖Before classASYNC

~55 min

#	Subtopic	Type	Time	What to do
1	Parity testing — what the agent can't do	Reading	15 min	Map every flow: which can Claude test, which can only a human check? Tests passing ≠ product working.
2	The first-user-path test	Reading	10 min	Run one test with a completely fresh account — no pre-seeded data. Hides what real new users hit.
3	Quiz — answer before class	Quiz	10 min	5 questions on the readings.
4	Instructions — how to prepare	Instructions	20 min	Feature deployed. Skill written. At least 2 compounding entries. Pattern scan written: what did you repeat manually that Claude could have run?

🎥In classLIVE

53 min + ~22 min FAQ

#	Subtopic	Type	Time	What happens
1	Recap of reading	—	5 min	"You read that tests passing ≠ product working. Today you watch a project fail the parity test."
2	Parity testing	Demo	15 min	Instructor takes a real participant project. Maps every flow. Identifies what Claude cannot test. Shows the gap.
3	The first-user-path test	Demo	10 min	Instructor runs a fresh-account test — no pre-seeded data. Shows what breaks.
4	Hook conversion optional	Demo	8 min	For participants who wrote skills and rules in Week 3: one rule-to-hook conversion. "Next step when you're ready." Not required.
5	Graduation	Exercise	20 min	Each participant presents 3 artifacts (see graduation bar). Group discussion.
—	FAQ	—	~22 min	—

✏️After classASYNC

~40 min

#	Exercise	Type	Time	What to do
1	Ship and verify	Exercise	20 min	Deploy your feature. Walk the verification checklist: URL, flows, console.
2	Pattern scan final output	Exercise	20 min	Review sessions from the cohort. What should become a skill or hook? One commitment for next week. Share before graduation.

Graduation bar — v1.5 · No hooks required

Artifact 1

One shipped feature

With a test that verifies it works. Proves you can execute end to end.

Artifact 2

One parity test

One flow in your project that Claude can't test automatically + your manual check for it. Proves you understand the limits.

Artifact 3

One compounding entry read aloud

What happened / why / what you'll do differently. Proves the system learned.

Bonus — not gated

One custom skill or hook

Celebrated but not required for graduation. Hooks removed from graduation requirement — developer-adjacent infrastructure, abstract for non-technical operators at this stage.