> ## Documentation Index > Fetch the complete documentation index at: https://onetest.ai/llms.txt > Use this file to discover all available pages before exploring further. # AI Best Practices > Get the most out of the OneTest AI Assistant Learn how to write better prompts, get more accurate results, and use the AI Assistant effectively. ## Writing Better Prompts ❌ "Create a test" ✅ "Create a test for login with OAuth via Google on mobile" ❌ "Generate tests" ✅ "Generate smoke tests for the checkout flow in our e-commerce app" ❌ "Test the API" ✅ "Test the POST /users API endpoint with valid and invalid email formats" ❌ "Create tests for signup" ✅ "Create 3 separate tests for signup: happy path, invalid email, and weak password" ## Iteration Techniques Begin with a general request: ``` Create tests for user authentication ``` Ask the AI to adjust: ``` Add steps for two-factor authentication Include social login scenarios Make the test data more specific ``` Modify scope as needed: ``` Split this into separate tests for each auth method ``` ``` Expand this to cover all error scenarios ``` Once you're happy with the format, use it as a reference: ``` Create more tests like TC-0042 ``` ## Do's and Don'ts Provide examples of what you want Mention your product type or industry Specify test priorities and tags Ask for explanations if unclear Request test data to be included Mention platform (web, mobile, API) Reference user stories or requirements Ask the AI to review existing tests Don't assume the AI knows your internal terminology Don't skip reviewing AI-generated content Don't use vague requests without context Don't forget to specify test type (manual/automated) Don't include sensitive production data in prompts Don't expect perfect results on first try ## Advanced Techniques Upload or paste requirements directly: ``` Generate tests from this user story: [paste entire user story with acceptance criteria] ``` Upload API documentation: ``` Create API tests from these Swagger docs: [attach or paste OpenAPI spec] ``` Use existing tests as templates: ``` Create tests similar to TC-0042 but for the signup flow ``` ``` Update TC-0156 to include these new requirements: [paste requirements] ``` Get suggestions: ``` What edge cases should I test for file upload? ``` ``` Review my checkout tests and suggest improvements ``` ``` What tests am I missing for this feature? ``` Work with multiple tests: ``` Generate a complete test suite for user management including CRUD operations ``` ``` Create tests for all error scenarios in the payment flow ``` ``` Convert these manual tests to automated test scripts ``` ## Common Patterns ### Pattern 1: Feature-Complete Testing ``` Generate a complete test suite for [feature] including: - Happy path scenarios - Error handling - Edge cases - Performance requirements - Accessibility checks ``` ### Pattern 2: From User Stories ``` Create tests for this user story: Title: As a [user type], I want to [action] so that [benefit] Acceptance Criteria: - [criterion 1] - [criterion 2] - [criterion 3] ``` ### Pattern 3: API Testing ``` Generate API tests for [endpoint] including: - Valid request - Invalid parameters - Authentication failures - Rate limiting - Response validation ``` ### Pattern 4: Regression Suite ``` Create regression tests for [feature] covering: - Previously reported bugs - Recent changes - Integration points - Critical workflows ``` ## Quality Checklist Before saving AI-generated tests, verify: Title clearly describes what is being tested Steps are specific and actionable Expected results are measurable Preconditions are listed Test data is appropriate Tags and priority are correct Test is atomic (tests one thing) No assumptions about system state ## Troubleshooting **Try:** * Rephrase your question * Provide more context * Give an example of what you want * Break complex requests into smaller parts **Try:** * Include specific test data * Mention your product/industry * Reference existing tests as templates * Provide acceptance criteria or requirements **Try:** * Show an example of your preferred format * Specify which fields to include * Ask for specific sections to be expanded * Request step-by-step refinements **Try:** ``` What edge cases am I missing for [feature]? ``` ``` Generate negative test scenarios for [feature] ``` ``` What error conditions should I test? ``` ## Pro Tips **Save Good Prompts:** When you craft a prompt that works well, save it for future use **Use Templates:** Once you have a test format you like, ask the AI to create more tests using the same structure **Iterate in Conversation:** Don't try to get perfect results in one prompt—refine through conversation **Combine with OQL:** Use the AI to help build complex OQL queries for advanced searches ## What's Next? Put these practices to use Master advanced searches