browser-load-test

Playwright browser load testing finds real concurrent agent limits by simulating realistic user interaction through the dashboard UI and API. It applies to capacity planning, scaling validation, and debugging dispatch behavior for complex multi-agent workloads.

Run the monitor in one terminal and execute the load test script in another to submit tasks via UI and API. Scripted phases of warmup, teams, stress, and cooldown poll status until saturation stops, outputting screenshots and time-series metrics JSON.

Playwright load testing generates screenshots and a time-series metrics JSON file for post-analysis. These artifacts capture evidence of dashboard behavior during warmup, peak stress, and cooldown phases to understand scaling limits and dispatch routing.

Yes, you need Playwright and httpx dependencies to perform browser-based load testing. Playwright drives browser UI flows simulating realistic user interaction, while httpx handles API task submissions required to saturate dashboard queues.

Yes, you can use Playwright load testing for debugging multi-agent dispatch behavior. It applies to debugging dispatch behavior for complex multi-agent workloads by submitting tasks and capturing metrics time-series JSON evidence during stress phases.

Dashboard dispatch behavior breaks under high concurrency when real concurrent agent limits are exceeded. Load testing identifies this saturation point by polling task status until scaling stops, helping tune dispatch routing and rate limits to prevent failures.