Claude Code Testing Workflow

Updated May 2026 · TDD, Test Generation & Debugging · Python, TypeScript, Go

Claude Code reads your entire codebase, understands your existing test patterns, and can write, fix, and debug tests fluently. This page covers the core testing workflows: generating tests for existing code, TDD (writing tests first), and debugging failing tests.

Quick Commands Reference

# Generate tests for a file
claude "write tests for src/auth.ts"

# TDD: make failing tests pass
claude "make the failing tests in auth.test.ts pass without changing the test file"

# Debug a failing test
claude "these tests are failing: $(npm test 2>&1 | tail -40). Fix the root cause."

# Generate tests for all untested files in a directory
claude "scan src/utils/ and generate test files for any functions that don't have tests yet"

# Add edge cases to existing tests
claude "look at src/auth.test.ts and add edge cases for null inputs, empty strings, and concurrent calls"

# Check test coverage and suggest what's missing
claude "review the test coverage for src/payments/ and list the most important missing test cases"

TDD Workflow with Claude Code

The red-green-refactor loop works well with Claude Code because it holds your full codebase in context and won't over-engineer implementations just to pass tests.

claude "write a failing Jest test for a validateEmail function that:
- returns true for valid emails
- returns false for missing @ symbol
- returns false for domains without a TLD
- returns false for null/undefined input
Don't implement the function yet."

npm test -- --testPathPattern=auth.test.ts
# → FAIL: validateEmail is not defined

claude "implement validateEmail in src/auth.ts to make the failing tests in src/auth.test.ts pass. Don't change the test file."

claude "refactor validateEmail to use a compiled regex for performance. Tests must still pass."

Generating Tests for Existing Code

TypeScript / Jest

# Basic: generate tests for one file
claude "write Jest tests for src/stripe-webhook.ts. Match the test style in src/__tests__/ (using describe/it/expect). Cover happy path, invalid signatures, and Stripe retry events."

# Claude will:
# 1. Read src/stripe-webhook.ts
# 2. Read existing tests in src/__tests__/ for style patterns
# 3. Generate src/__tests__/stripe-webhook.test.ts
# 4. Run the tests and fix any that fail

Python / Pytest

# Generate pytest tests for a module
claude "generate pytest tests for src/auth/validators.py. Use fixtures for database setup. Cover validation errors with pytest.raises. Mock external HTTP calls with responses library."

# For a Django view:
claude "write tests for the UserProfileView in accounts/views.py using Django's TestCase. Cover: GET returns 200 for authenticated users, GET returns 403 for anonymous, POST updates profile, POST with invalid data returns 400 with errors."

Go

# Generate Go table-driven tests
claude "write Go tests for the ParseConfig function in config/parser.go. Use table-driven test style (var tests = []struct{...}). Cover: valid YAML, invalid YAML, missing required fields, default values."

# Claude generates standard Go test file:
# func TestParseConfig(t *testing.T) {
#   tests := []struct { name string; input string; want Config; wantErr bool }{...}
#   for _, tt := range tests { ... }
# }

Debugging Failing Tests

Paste the error output

# Pipe test output directly to Claude
npm test 2>&1 | claude --no-interactive "these tests are failing. Identify the root cause and fix it without changing test assertions."

# Or paste interactively
claude
> these tests failed:
> FAIL src/auth.test.ts
>   ● validateEmail › should return false for missing @ symbol
>     Expected: false
>     Received: true
>
> Fix the root cause in the source file, not the test.

Flaky test diagnosis

claude "src/payments.test.ts fails intermittently — about 1 in 10 runs — with a timeout error on the Stripe webhook handler. It's consistent in CI but not locally. What are the likely causes and how do I fix them?"

Claude typically identifies race conditions, uncleared mock state, clock drift in time-dependent assertions, or missing await on async teardown.

Coverage gap analysis

# After running coverage
npm test -- --coverage 2>&1 | claude --no-interactive "which functions have the lowest coverage? What are the most important missing test cases for the payment processing logic?"

Setting Up CLAUDE.md for Consistent Test Generation

Add a testing section to your CLAUDE.md so Claude always generates tests in your preferred style without being told:

# CLAUDE.md (in project root)

## Testing Conventions
- Framework: Jest + React Testing Library
- Test files: colocate with source as `*.test.ts`
- Use `describe()` blocks per exported function
- Use `it()` for each case (not `test()`)
- Mock external APIs with `jest.mock()` at module level
- Database: use in-memory SQLite for tests (see `src/test-utils/db.ts`)
- Never use `any` in test assertions — use proper types
- Always test error paths and null inputs

With this in CLAUDE.md, claude "write tests for src/auth.ts" will automatically follow all these rules without you repeating them.

Common Patterns and Pitfalls

Frequently Asked Questions

→ Claude Code debugging workflow

→ Automate testing in CI/CD with Claude Code

→ Code review workflow

← Back to Home