# TestLink MCP for AI Agents AI Agent Connect > TestLink connects your self-hosted test management instance to your AI agent. It lets you query test projects, plans, suites, and specific test cases using natural language instead of clicking through a legacy UI. You can pull manual test steps, check build versions, and verify preconditions directly in your chat or code editor. It bridges the gap between your classic QA data and modern development workflows. ## Overview - **Category:** developer-tools - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_Txd0bdinEzyETBUMcUstyIr6u481CAjFsmC77a1k/ai-agent-connect - **Tags:** qa-testing, test-management, test-execution, software-quality, bug-tracking, test-plans ## Description You spend half your day digging through old test management folders just to find the right steps for a bug fix or a regression cycle. It's a chore to jump between your code editor and a separate tab full of test cases, especially when you're trying to stay in the zone. This Connector changes that by letting your AI agent do the heavy lifting. Instead of hunting for a specific test plan or checking which build is currently assigned to a project, you just ask. Your agent pulls that data instantly, giving you the exact preconditions or manual steps you need right where you're already working. You can see the hierarchical structure of your test suites or grab a list of active builds without ever leaving your prompt. It turns a fragmented QA process into a conversational one. By adding this to the Vinkius catalog, you get a way to make your self-hosted TestLink instance actually talk to the modern tools you use every day. It's about getting the right information at the right time without the manual overhead of navigating a complex, older interface. ## Tools ### get_test_plan_details Shows the specific details of a test plan. Use this to see what a plan covers at a glance. ### get_test_project_details Retrieves the core information for a TestLink project. It's the starting point for finding any other project data. ### get_test_suite_details Gets the specific data for a test suite. This is useful for understanding a specific branch of your test coverage. ### list_test_builds Shows all builds associated with a test plan. Use it to confirm which version of the software is being tested. ### list_test_cases Lists every test case inside a suite. This is great for getting a high-level view of what's being tested. ### list_test_plans Shows all plans for a specific project. Use this to see the full roadmap of testing for a project. ### list_test_projects Lists every project on your TestLink server. It gives your agent a bird's eye view of your entire testing environment. ### list_test_suites Shows all suites within a project. This helps you navigate the hierarchy of your test organization. ### get_user_info Shows who is currently logged in. This helps you confirm your agent has the right permissions to see the data. ### get_test_case_details Pulls the full instructions and expected results for a specific test case. It helps you see exactly what to do without opening a browser. ## Prompt Examples **Prompt:** ``` Who am I logged in as and can I see the projects? ``` **Response:** ``` You are authenticated as `QA_Admin` on project footprint ID `1`. I can see the following projects: [Project A, Project B, Project C]. How would you like to proceed? ``` **Prompt:** ``` What are the steps for Test Case 5440? ``` **Response:** ``` Here are the details for Test Case 'Billing Module Crash' (ID 5440): **Preconditions**: User must have a cart with exactly $0 in amount. **Step 1**: Attempt to proceed to checkout bypassing frontend validation using cURL. (Expected: 400 Bad Request error returned). **Step 2**: Open Stripe gateway page. (Expected: Page throws exception cleanly). ``` **Prompt:** ``` Show me the plans for Project 2. ``` **Response:** ``` I found 2 active test plans mapped directly to Project ID 2: 1. **Release v1.5 - Regression Sandbox** (Plan ID: 1045) 2. **Payment Gateway E2E** (Plan ID: 1993) Which plan would you like to retrieve the registered builds for? ``` ## Capabilities ### Pull test project details Get the status and metadata of any project in your TestLink instance. ### Browse test plans See all plans linked to a specific project to identify the right scope. ### List test suites View the folder structure of your test coverage for a project. ### Read test case steps Get the full manual instructions, expected results, and priorities for a case. ### Check active builds See which software builds are assigned to a specific test plan. ### Verify user identity Confirm which user your agent is currently authenticated as. ## Use Cases ### Defining a regression cycle A QA Lead asks the agent to list all plans for a project to see what needs to be included in the next release. ### Documenting a bug fix A developer asks for the manual steps of a specific test case to ensure they understand the expected behavior of a feature. ### Verifying build versions An engineer asks which builds are active for a specific plan to make sure they are testing the correct software version. ### Auditing test coverage A manager asks the agent to list all suites in a project to see how much of the application is currently covered by tests. ## Benefits - Stop clicking through nested folders. Use list_test_projects and list_test_suites to let your agent navigate the hierarchy for you. - Get manual steps instantly. Use get_test_case_details to pull full instructions into your editor without leaving your current workspace. - Confirm QA targets faster. Use list_test_builds to see which software versions are linked to your active test plans immediately. - Simplify regression planning. Use list_test_plans to pull all available testing scopes into a single conversation for faster decision making. - Remove the QA bottleneck. Let developers use get_test_case_details to see preconditions themselves instead of waiting for a teammate to reply. ## How It Works The bottom line is you get instant access to your QA test data through a natural language chat. 1. Subscribe to the TestLink MCP on Vinkius. 2. Provide your instance Base URL and Developer API Key in your client settings. 3. Ask your agent to find test plans, check build versions, or read test steps. ## Frequently Asked Questions **Can the TestLink MCP access my self-hosted data?** Yes, it connects directly to your self-hosted TestLink instance using your Base URL and Developer API Key. **Can I use this to read manual test steps?** Yes, you can ask your agent to pull the full manual instructions and expected results for any specific test case. **Does this work with Cursor or Claude?** It works with any MCP-compatible client, including Claude, Cursor, and Windsurf. **How do I check which software build is assigned to a plan?** You can ask your agent to list the builds for a specific test plan, and it will return the target versions. **Can it help with regression planning?** It can list all test plans for a project, making it much faster to see what needs to be covered in a new cycle. **Is my data secure?** The Connector uses your own Developer API Key to authenticate, so your data stays within your environment. **Why should I plug an AI agent into an older legacy application like TestLink?** Because retrieving test assets from older GUIs can be painfully slow for developers. By mapping the database to an AI agent, your devs can just chat: 'What are the steps to reproduce Test Case 4022?' The agent calls the XML-RPC backend and instantly formats the manual steps logically in the chat space. **Can it read all my hierarchical QA Suites and specific cases?** Yes! The testlink agent is configured to retrieve Suites. You just list a Project ID, traverse downwards into Test Suites, and then request Test Cases by the retrieved Suite ID. It allows full architectural transparency over chat. **Does my TestLink server need to be accessible to the internet?** No. Because Connectors process logic strictly locally through a sidecar setup, the interaction happens right on your machine/endpoint. As long as the machine running the AI client (e.g. Cursor, Claude Desktop) can resolve the Internal Network URL, the connection is solid.