Start Small
What this lesson teaches
The best way to learn Claude Code is through real tasks—but not huge ones. This lesson covers how to pick the right starting tasks and work through them effectively.
Why Start Small?
Big tasks have too many variables. When something goes wrong, you won't know if it's your prompt, Claude's interpretation, or the task complexity. Small tasks give you:
- Clear feedback loops
- Lower risk of wasted effort
- Faster learning through more iterations
Good First Tasks
Bug Fixes
Start with bugs that have clear reproduction steps. Tell Claude:
> There's a bug where clicking "Save" twice creates duplicate entries.
> The handler is in src/handlers/save.ts. Fix it.
Adding Small Functions
Functions with clear inputs and outputs are perfect:
> Add a function to utils/format.ts that converts dates
> to "2 days ago" style relative time.
Simple Refactoring
Extract functions, rename variables, or simplify logic:
> The validateUser function in auth.ts is 80 lines.
> Break it into smaller functions.
Pattern: Good first tasks are isolated (don't touch many files), have clear success criteria, and can be verified quickly.
Working Through a Task
- State the goal clearly — What should happen when you're done?
- Point to relevant files — Don't make Claude search
- Review the changes — Read diffs before accepting
- Test immediately — Run the code, check the fix
Common mistake: Accepting changes without understanding them. If you can't explain what changed, you'll struggle to debug issues later.
Key Takeaways
- Start with isolated, verifiable tasks
- Bug fixes and small functions are ideal starters
- Always understand changes before accepting
- Test immediately after each change