How to Use Claude Code with Jest Testing

Written by Michael Lip · Solo founder of Zovo · $400K+ on Upwork · 100% JSS Join 50+ builders · More at zovo.one

The Problem

Writing and maintaining Jest tests is time-consuming. You want Claude Code to write tests, run them, analyze failures, and fix the code or tests automatically, but getting reliable results requires the right workflow.

Quick Fix

Pre-approve Jest commands in your project settings and ask Claude to implement with tests:

{
  "permissions": {
    "allow": [
      "Bash(npx jest *)",
      "Bash(npm run test *)",
      "Bash(pnpm test *)"
    ]
  }
}

Then: “Write tests for src/api/handlers/users.ts and run them”

What’s Happening

Claude Code can run Jest directly through its Bash tool, read test output, analyze failures, and iterate on fixes. The key to reliable results is pre-approving test commands so Claude can run tests without permission prompts at every step, and structuring your CLAUDE.md with test conventions so Claude writes tests that match your project’s patterns.

Step-by-Step Fix

Step 1: Configure test permissions

Add test-related commands to your project settings at .claude/settings.json:

{
  "permissions": {
    "allow": [
      "Bash(npx jest *)",
      "Bash(npm run test *)",
      "Bash(pnpm test *)",
      "Bash(npx jest --coverage *)"
    ]
  }
}

Step 2: Add test conventions to CLAUDE.md

Tell Claude how your project handles testing:

# Testing
- Test framework: Jest with TypeScript
- Test files: `*.test.ts` next to source files
- Run all tests: `pnpm test`
- Run single test: `pnpm test -- --testPathPattern=path/to/test`
- Coverage: `pnpm test -- --coverage`
- Mock conventions: Use `jest.mock()` for external dependencies
- Test structure: describe/it blocks, arrange-act-assert pattern
- Always test error cases and edge cases, not just happy paths

Step 3: TDD workflow with Claude Code

Ask Claude to write tests first, then implement:

Write failing Jest tests for a UserService.create() method that:
- Creates a user with valid input
- Throws ValidationError for missing email
- Throws DuplicateError if email already exists
- Hashes the password before storing

Then implement the method to make all tests pass.

Claude will:

  1. Write the test file
  2. Run the tests (they should fail)
  3. Implement the method
  4. Run the tests again (they should pass)
  5. Iterate if any tests still fail

Step 4: Auto-run tests after edits with hooks

Configure a PostToolUse hook that runs related tests after every file edit:

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "FILE=$(jq -r '.tool_input.file_path'); case \"$FILE\" in *.ts|*.tsx) npx jest --findRelatedTests \"$FILE\" --passWithNoTests 2>&1 | tail -20 ;; esac"
          }
        ]
      }
    ]
  }
}

This runs jest --findRelatedTests on the edited file, so Claude immediately sees test results after every edit.

Step 5: Debug failing tests

When tests fail, give Claude the full context:

Run pnpm test and fix any failures. Show me what was wrong and what you changed.

Claude reads the test output, identifies the root cause, and fixes either the implementation or the test.

Step 6: Generate coverage reports

Run pnpm test -- --coverage and identify files under 80% coverage.
Write tests for the uncovered paths.

Step 7: Reduce test output context

Large test outputs fill the context window. Use a hook to filter to failures only:

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "if": "Bash(npx jest*)",
            "command": "jq -r '.tool_output' | grep -A 5 'FAIL\\|Error\\|expected\\|received' || echo 'All tests passed'"
          }
        ]
      }
    ]
  }
}

Prevention

Add testing conventions to your CLAUDE.md so Claude writes tests consistently. Pre-approve test commands to eliminate permission prompt friction. Use the --findRelatedTests flag in hooks to keep test runs fast and focused.

Start each testing task with /clear to avoid stale context from previous work contaminating the test analysis.

Level Up Your Claude Code Workflow

The developers who get the most out of Claude Code aren’t just fixing errors — they’re running multi-agent pipelines, using battle-tested CLAUDE.md templates, and shipping with production-grade operating principles.

Get Claude Code Mastery — included in Zovo Lifetime →

16 CLAUDE.md templates · 80+ prompts · orchestration configs · workflow playbooks. $99 once, free forever.