Demo
A coding agent sets up API flow tests, CI flags a slower flow, @stresseur explains why, and a cloud load test finds the limit.
Transcript
In the AI era, teams ship so fast that their APIs slow down. Start with your coding agent. Ask it to set up API flow tests. It installs the Stresseur skill and links your repo to your workspace. Then the Stresseur recorder records real web usage. The agent starts your app and uses the shop in a browser. It logs in, adds a category, creates a product, and edits the price. Each API call is counted as it happens and nothing leaves your machine. The session becomes a YAML flow. The agent runs it. The login step fails, so it finds the test password and runs it again. Every step passes. Next, it adds a GitHub Actions workflow. Every pull request runs each flow thirteen times. The agent commits it on a branch. In the app, the overview shows what needs attention first. Admin catalog CRUD got slower on master. Each run is compared with recent master runs on the same kind of CPU. So a slow CI machine won't wake you up. The flow page names the commit and the steps that got slower. Every pull request gets a check and one comment. It says which flows got slower, by how much, and if the gate fails. Ask it: “@stresseur why is this slower?” It reads the code change and the stored runs. It answers with evidence, and every number links to its run. Need to know how much traffic you can take? Run the same flows as a cloud load test. It shows the point where your p95 goes over your limit. Ask your coding agent to set it up.