How to use the Integration Testing Guide agent skill

What the integration-testing skill makes the agent do

Once installed, the agent looks for the points where parts of your system meet: API routes calling services, services interacting with the database, and components reading shared state. It checks whether those interactions are tested, suggests or writes integration tests for the critical paths, and flags gaps where CI does not automatically catch regressions.

The skill also explains how integration tests differ from unit and end-to-end tests. Unit tests check individual functions in isolation, but they can miss bugs that appear when those functions are wired together. Integration tests are meant to catch those wiring bugs before they reach production.

• Finds integration points in the codebase
• Checks whether those interactions have tests
• Suggests or writes integration tests for critical paths
• Flags where failures do not block regressions in CI
• Explains unit, integration and end-to-end testing

Install the integration-testing skill

Download the ZIP for this skill from this page and unzip it. The skill folder contains SKILL.md and references/rule.md. Put the unzipped folder in one of these locations:

You can install it for yourself across projects by placing it in ~/.claude/skills/, or for a single project by placing it in that project's .claude/skills/ folder.

Alternatively, in the Claude apps, go to Settings → Capabilities → Skills and upload the skill there.

• Personal: ~/.claude/skills/integration-testing/
• Project only: .claude/skills/integration-testing/ inside your repository
• Claude apps: Settings → Capabilities → Skills, then upload

How to trigger the skill with natural requests

You do not need a special command. The skill's description tells the agent to use it when you ask about CI coverage, automated checks, or test strategy for integration tests. Plain requests work well, for example asking whether your key workflows are covered by integration tests, or whether CI actually enforces your test strategy.

Be specific about the area you care about. Naming a workflow such as user sign-up or checkout gives the agent a clear target.

• Are the critical paths in our API routes covered by integration tests?
• Check whether CI would catch a regression in our checkout flow.
• Review where our services touch the database and tell me which interactions have no tests.
• Explain how integration tests differ from unit and end-to-end tests in this project.

Examples of what the agent produces

For a typical review, the agent identifies the integration points, lists which ones have tests and which do not, and proposes tests for the most important paths. The skill's guidance is to use a real database in tests (a test database or in-memory database) rather than mocking everything, and to cover happy paths and the most critical error paths.

The reference material includes a code example of an integration test for a POST /api/users route that migrates and seeds a database, sends a request with supertest, and checks the response body. The agent can adapt this pattern to your framework.

What you need before you start

The skill lists no required tools. However, it assumes your project already has some testing infrastructure and CI setup to review. Without a test suite or CI configuration, there is little for the agent to check.

The output is code and test configuration. Someone with engineering judgment should review it before applying it to a production project.

• A codebase with a test suite and CI setup
• Engineering review of any generated tests before merging

Safety notes: scripts and network access

This skill does not run scripts and does not use the network, according to its listing. It works by reading your code and producing suggestions and test code. Review any changes the agent makes to test files and CI configuration before committing them.

• Runs scripts: no
• Uses network: no

Tips for better results

Give the agent a specific workflow to focus on rather than the whole repository, so the review stays concrete. Ask it to flag where failures would not block a merge, since the skill specifically looks for gaps where CI does not enforce the rule. If you are new to testing, ask for an explanation of the difference between unit, integration and end-to-end tests first.

• Name one workflow at a time
• Ask specifically whether CI blocks regressions
• Ask for explanations alongside the review if you are learning