# Cypress Cloud MCP for AI Agents AI Agent Connect > Cypress Cloud MCP. Audit E2E test results via your AI agent. Monitor test runs, inspect failed spec instances for screenshots and video URLs, track flaky tests, and pull performance metrics like p95 durations directly from your Cypress Cloud Enterprise account. ## Overview - **Category:** loved-by-devs - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_IKhVCeL6avFJPY6iQ838105q3zZBtwMa7WMfyaRI/ai-agent-connect - **Tags:** e2e-testing, test-automation, quality-assurance, debugging, test-reporting, ci-cd-pipeline ## Description Managing end-to-end testing at scale usually means jumping between your CI/CD pipeline and the Cypress Cloud dashboard. You spend half your time digging through logs to find out why a specific test failed in a specific environment. This Connector changes that by giving your AI agent direct access to your test data. Instead of hunting for a specific run, you can just ask your agent to show you the last failed tests or pull the error messages for a specific spec. It pulls in the screenshots, video URLs, and the exact commit info so you can see exactly what went wrong without leaving your workspace. If you're trying to figure out why your pipeline is slowing down, you can ask for the average durations or the p95 metrics for your slowest tests. It's a way to turn your test results into a conversation. You'll find it listed in the Vinkius catalog, where it connects your enterprise account to your favorite AI client. It's about getting answers faster so you can spend less time on data entry and more time actually fixing bugs. ## Tools ### get_instances List every spec file execution within a specific run. It returns instance IDs, spec names, and durations. ### get_instance Get full details for a single spec instance. This includes the error message, screenshot, and video URL. ### get_tests List individual tests within a spec instance. It shows titles, states, durations, and error messages. ### report_runs Get an enterprise run summary report. This is great for pulling aggregated data into BI dashboards. ### report_tests Get enterprise test results reports. It provides individual test-level data with statuses and error messages. ### report_flaky Get a report of flaky tests. It identifies tests that pass and fail intermittently and lists their flake rates. ### list_projects Find your project names and IDs across your Cypress Cloud organization. Use this to get the correct project ID for other queries. ### get_runs List recent test runs for a project. It shows you the status, branch, duration, and commit info for each run. ### get_run Get full details for a specific run. This includes passed and failed counts, parallelization, and group info. ### report_slow Get a report on your slowest tests. It returns test names along with average, p95, and max durations. ## Prompt Examples **Prompt:** ``` Show me the last 3 failed runs for project 'abc123'. ``` **Response:** ``` I found 3 recent failed runs for project 'abc123': | Run ID | Branch | Status | Duration | Commit | | :--- | :--- | :--- | :--- | :--- | | 102 | main | Failed | 45s | a1b2c3d | | 98 | dev | Failed | 12s | e5f6g7h | | 95 | main | Failed | 38s | i9j0k1l | Would you like me to pull the error messages for the latest one? ``` **Prompt:** ``` Why did the 'Checkout' test fail in the last run? ``` **Response:** ``` The 'Checkout' test failed in the latest run. Here are the details: * **Error:** `Timed out retrying after 4000ms: Expected to find element: .success-message` * **Duration:** 4.2s * **Screenshot:** [View Screenshot](https://cypress.io/example-screenshot.png) * **Video:** [Watch Recording](https://cypress.io/example-video.mp4) It looks like the success message didn't appear in time. ``` **Prompt:** ``` Give me a report of the slowest tests from the last week. ``` **Response:** ``` I've analyzed the performance metrics for the past week. Here are your top 3 slowest tests: 1. **User Onboarding Flow** - Avg: 58s, p95: 62s 2. **Payment Gateway Sync** - Avg: 42s, p95: 45s 3. **Bulk Image Upload** - Avg: 35s, p95: 39s Would you like me to pull a full report for your dashboard? ``` ## Capabilities ### List recent test executions Get a list of recent test runs including status, branch info, and commit IDs. ### Inspect failed spec details Retrieve error messages, screenshots, and video URLs for specific failed tests. ### Identify flaky tests Find tests that intermittently pass or fail and see their last flake dates. ### Audit performance metrics Pull slow test reports to see average and p95 durations for your CI/CD pipeline. ### Generate enterprise reports Fetch aggregated run summaries and test results for your BI dashboards. ### Navigate projects List all organizational projects and find the unique IDs needed for data extraction. ## Use Cases ### Debugging a failed login flow A QA engineer asks the agent to find the last failed run for the 'auth' project and pull the screenshot and error message for the login spec. ### Identifying flaky tests in the pipeline A DevOps engineer asks the agent to list all tests with a flake rate higher than 5% to determine which specs need refactoring. ### Reporting on pipeline performance A manager asks the agent to fetch a report of the top 10 slowest tests from the last week to identify bottlenecks in the CI/CD flow. ### Checking commit-specific test results A developer asks the agent to list the test results for a specific commit ID to see if the new changes broke any existing specs. ## Benefits - Get screenshots and video URLs instantly using get_instance to see exactly what happened during a failure. - Spot flaky tests before they break your build by using report_flaky to see flake rates and last flake dates. - Identify performance bottlenecks in your CI/CD pipeline with report_slow p95 duration metrics. - Pull aggregated data for your BI tools using report_runs to keep stakeholders updated on quality metrics. - Quickly find specific project IDs across your organization using list_projects to keep your workflow moving. - Debug faster by asking your agent to pull error messages for specific spec instances directly in your workspace. ## How It Works The bottom line is you get a conversational interface for your entire Cypress Cloud test history. 1. Subscribe to the Connector and enter your Cypress Cloud Enterprise API Key and Project ID. 2. Connect the Connector to your AI client like Claude, Cursor, or Windsurf. 3. Ask your agent to audit test results, find flaky tests, or pull performance reports. ## Frequently Asked Questions **Can I use Cypress Cloud MCP with my existing Enterprise account?** Yes, this Connector connects directly to your existing Cypress Cloud Enterprise account using your API key and project IDs. **Does Cypress Cloud MCP show me screenshots of failed tests?** Yes, it can retrieve specific screenshots and video URLs for failed spec instances so you can see exactly what went wrong. **How does Cypress Cloud MCP help with flaky tests?** It identifies tests that intermittently pass or fail by pulling flake rates and the dates of the last occurrences, helping you prioritize what to fix. **Can I get performance metrics like p95 with Cypress Cloud MCP?** Yes, you can request reports on slow tests that include average, p95, and max durations for your CI/CD pipeline. **Is Cypress Cloud MCP compatible with Cursor and Claude?** Yes, it works with any MCP-compatible client, including Claude, Cursor, Windsurf, and VS Code. **Do I need a special license for Cypress Cloud MCP?** You need a Cypress Cloud Enterprise account to use this Connector, as it utilizes the Enterprise Data Extract API. **Can my agent help me debug a failed Cypress test?** Yes. Use the 'get_instance' tool with the instance ID. The agent will retrieve explicit error messages and can provide you with direct links to failure screenshots and video recordings stored in Cypress Cloud. **How do I identify which tests are flaky in my project?** Provide a start date to the 'report_flaky' tool. Your agent will analyze the historical data from Cypress Cloud and identify tests that intermittently pass and fail, helping you improve your CI reliability. **Can I use the agent to monitor the duration of my test runs?** Absolutely. The 'report_slow' tool retrieves high-precision duration metrics. The agent can surface the slowest tests by average or p95 duration, allowing you to optimize your test suite's execution time.