# LoadNinja (Real-Browser Load Testing) MCP > LoadNinja (Real-Browser Load Testing) MCP gives your AI agent full control over performance engineering. Manage complex load tests—from listing saved scenarios to triggering live, high-volume traffic runs—all through natural conversation. Get detailed metrics on throughput, peak user counts, and transaction times without ever touching a dashboard. ## Overview - **Category:** developer-tools - **Price:** Free - **Tags:** load-testing, performance-engineering, browser-automation, scalability, stress-testing ## Description Need to know if your website handles 10,000 users at checkout? Instead of spending hours navigating dashboards and manually exporting CSVs, you just talk to your agent. This MCP connects your AI client directly to LoadNinja’s performance infrastructure. You can ask it to list all existing load scenarios or trigger a new test run instantly, specifying the virtual user count and duration in minutes. After the test finishes, you don't sift through raw data; you simply ask for an analysis of throughput, peak usage, and transaction timing. The agent handles the complexity, giving you actionable performance insights right where you work. By connecting this MCP via Vinkius, your entire team gains a single point of access to manage all load testing needs, whether you're running basic smoke tests or full-scale stress simulations. ## Tools ### list_scenarios Shows you every saved load test scenario in your account. ### get_scenario Retrieves the full configuration details, including target URLs, for one specific load test scenario. ### run_scenario Starts a new performance run by executing a saved scenario with specified virtual users and duration in minutes. ### list_test_runs Lists all previous test executions, showing their current status (running or completed). ### get_test_run Retrieves the basic details and summary information for a specific completed test run. ### get_test_run_stats Pulls the raw performance statistics, including server response times, for deep analysis of a single test run. ### stop_test_run Immediately and irreversibly cancels an active load test to prevent unnecessary resource usage. ### list_browsers Lists all explicit native browser types that can be used for load testing. ### list_locations Lists the physical data center locations available for global load injection testing. ### get_account Provides your account's current subscription details and strict maximum virtual user limits. ## Prompt Examples **Prompt:** ``` List all load test scenarios in my LoadNinja account ``` **Response:** ``` I've found 4 scenarios: 'Main-Store-Checkout' (ID: scen-123), 'User-Login-Flow' (ID: scen-456), 'Inventory-Search-API' (ID: scen-789), and 'Mobile-App-Smoke' (ID: scen-012). Which one would you like to run or inspect? ``` **Prompt:** ``` Run scenario 'scen-123' with 100 VUs for 10 minutes ``` **Response:** ``` Triggering load test scenario 'scen-123'… I've started the run with 100 virtual users for a duration of 10 minutes. New Run ID: 'run-98765'. I'll monitor the progress and notify you once the performance metrics are available. ``` **Prompt:** ``` Show me the performance stats for the last completed run ``` **Response:** ``` Retrieving metrics for run ID 'run-13579'… The test achieved a peak of 250 VUs with an average throughput of 45 requests/sec. The P95 transaction duration was 1.2s. No major bottlenecks detected. Would you like to see the raw metrics breakdown? ``` ## Capabilities ### List test scenarios Shows every saved performance test scenario you have set up in LoadNinja. ### Get specific scenario details Retrieves the full configuration and target URL for any single load test scenario. ### Run a live load test Starts a new performance run, specifying exactly how many virtual users and for what duration. ### List all past runs Shows the status of every test execution to help you track if it's completed or still running. ### Analyze run summaries Gathers high-level performance data, like average throughput and peak user counts, for finished tests. ### Get raw performance stats Extracts the deep, underlying metrics and server response times to pinpoint specific bottlenecks. ### Stop active test runs Immediately cancels a running load test to manage system resources or halt an error cycle. ## Use Cases ### Checking for holiday traffic readiness A QA tester needs to know if the checkout process can handle a major sales spike. They ask their agent: 'Run the Main-Store-Checkout scenario with 1000 virtual users for 30 minutes.' Once done, they use `get_test_run_stats` to verify that average throughput stayed above 50 requests/second. ### Debugging a regional outage A DevOps engineer notices latency spikes in Europe. They ask their agent to list all physical data center locations, see the European options, and then use `run_scenario` targeting that specific region to isolate the bottleneck. ### Verifying account capacity A team lead is preparing for a major launch. They first ask their agent to run `get_account` to verify their current subscription limits, ensuring they don't exceed the maximum virtual user count before starting any tests. ### Quickly reviewing last week’s metrics A performance engineer needs a quick summary of the previous day's run without pulling up historical dashboards. They ask for the summaries using `list_test_runs` and then request analysis via `get_test_run`. ## Benefits - Instant control: Instead of navigating complex dashboard menus, you simply tell the agent to 'Stop test run X.' This uses the `stop_test_run` tool, saving minutes of manual cleanup time. - Deep diagnostics: When a bottleneck pops up, don't guess. Use the `get_test_run_stats` tool to pull raw performance statistics and find exactly which server response time failed. - Global visibility: Need to know if you can scale internationally? The agent uses `list_locations` and `get_account` to show available physical data center points and your current VU limits instantly. - Full scenario management: Forget digging through folders. You use the `list_scenarios` tool to see every saved test configuration and `get_scenario` for its full details, all in one conversation. - Controlled execution: Running a new load test is simple. Just ask the agent to trigger the run with specified users using the `run_scenario` tool, and you're live immediately. ## How It Works The bottom line is that you manage complex, global-scale load testing by talking to your AI client instead of clicking through multiple dashboards. 1. Subscribe to this MCP and provide your LoadNinja API key credentials. 2. Your AI client uses these keys to connect directly to the LoadNinja performance platform. 3. You simply ask your agent a question, like 'Run the checkout scenario with 500 virtual users for 15 minutes,' and it executes the command. ## Frequently Asked Questions **How do I list all load test scenarios using LoadNinja (Real-Browser Load Testing) MCP?** You ask the agent to 'list all load test scenarios.' The tool uses `list_scenarios` to show you every saved performance configuration in your account. **Can I trigger a live test run with LoadNinja (Real-Browser Load Testing) MCP?** Yes. You tell the agent exactly which scenario, how many virtual users, and for what duration you want to test by calling `run_scenario`. **What is the difference between getting run details and raw stats with LoadNinja (Real-Browser Load Testing) MCP?** Use `get_test_run` for high-level summaries like overall success. Use `get_test_run_stats` when you need deep, raw performance statistics to pinpoint a specific bottleneck. **How do I check my account limits with LoadNinja (Real-Browser Load Testing) MCP?** The agent runs the `get_account` tool. This instantly verifies your current subscription details and strict maximum virtual user counts so you don't over-allocate resources. **If a test is running badly, how do I stop it with LoadNinja (Real-Browser Load Testing) MCP?** Simply ask the agent to terminate the run. It executes `stop_test_run`, which immediately and irreversibly vaporizes the active load cycle.