Step 3: The Daily Workflow (15 minutes)

The Core Loop

Every productive Claude Code session follows this rhythm:

Start → Request → Implement → Verify → Close

Here’s what that looks like in practice:

1. Start (1 minute)

cd your-project
claude

/check-env

Fix any issues. Then state your task.

2. Request (1 minute)

Write a clear, scoped request. Good format:

[What to do] in [which files/directory].
[Constraint]. [Verification criteria].

Good examples:

Fix the failing test in src/auth/login.test.ts.
The error is "Expected 200, received 401".
Only modify files in src/auth/. Don't change tests.

Add rate limiting to POST /api/login.
Max 5 attempts per minute per IP. Only modify src/auth/.
Ask me clarifying questions before you start.

Bad examples:

Fix the login                    ← too vague
Fix login, add dashboard, update docs  ← too many tasks
The code is broken               ← no error message

3. Implement (varies)

Claude reads files, plans the approach, and implements. Watch for:

Hooks firing — type checks, lint checks after each edit
Claude reading before writing — good sign it understands context
Scope staying focused — alert Claude if it starts touching unrelated files

4. Verify (1 minute)

Always ask Claude to prove the work:

Run npm test and tsc --noEmit. Show me the output.

Never accept “this should work” without actual test output.

5. Close

If switching tasks: /clear then start new request. If ending for the day: /handoff to save context for tomorrow.

The 5 Rules

Tape these to your monitor:

One task per session. Multi-task = lower quality.
Paste real errors. Don’t describe — paste.
Scope-lock prompts. “Only modify src/auth/”
Verify with real tests. “Run npm test, show output”
Fresh session above 80%. Use /handoff first.

When Things Go Wrong

Symptom	Fix
Claude giving generic answers	Context polluted → `/clear` or new session
Claude modifying wrong files	Add scope lock: “Only modify src/X/”
Claude over-engineering	Add: “Smallest change possible. No refactoring.”
Fix doesn’t work after 2 attempts	New session with clearer spec
“Works on my machine”	Run `/check-env` first

Next: 04-skills-tour.md — Hands-on tour of key skills