RequestPulse — Load Testing for VS Code

Apne server ki nabz pehchaano. Know your server's pulse before your users do.

Send a controlled wave of traffic to any local or deployed HTTP endpoint, and see exactly how it holds up — response times, errors and throughput — without ever leaving VS Code.

Works with localhost, Vercel, Render, or any URL you own. No accounts, no cloud service, no AI, no telemetry. Everything runs on your machine.

Why RequestPulse?

Most developers find out how their app behaves under load the worst possible way: launch day.

Proper load-testing tools are powerful, but they ask you to write scripts, read raw terminal output and work out what the numbers mean. RequestPulse puts the whole workflow inside your editor:

1. Type a URL, pick how much traffic and for how long.
2. Press Start.
3. Read a report that explains itself — in plain language, with a clear pass or fail against your targets.
4. Change your code, run it again, and compare the two runs side by side.

That last step is the point. It turns "I think caching made it faster" into "p95 went from 851 ms to 39 ms and errors dropped from 11% to 0%."

Features

Describe traffic the way you think about it

Choose requests per second ("20 per second for 5 minutes") or a total number of requests ("6,000 over 5 minutes"). RequestPulse spreads requests evenly across the duration, and the report shows the traffic actually delivered, because your computer may not keep up with high targets.

Set targets and get a verdict

Give each test two goals: a p95 response-time target and a maximum error rate. The report tells you whether the endpoint met them.

A report that explains itself

Every run produces:

• Requests sent, versus the rate you asked for
• Median, p90, p95, p99 and slowest response time
• Failed requests, with a breakdown of status codes, timeouts and connection errors
• "What this means" — plain-language findings, such as whether responses slowed down as the test went on, whether the server was rate-limiting you (HTTP 429), or whether the first requests looked like a cold start

Charts have hover tooltips, show your targets as reference lines, and use colour plus line style so they stay readable for everyone.

Live progress, and a Stop button that actually stops

Watch the progress bar, or stop at any moment. A stopped test keeps the results gathered so far and is clearly labelled as stopped, never passed off as a full run.

Save every run, then compare before and after

Each test is saved on your computer. Pick any two and get a side-by-side table, a one-sentence summary and overlaid charts. If the two runs used different settings, RequestPulse warns you, because that makes a comparison unfair.

Built to be safe

• Anything that is not your own computer requires you to confirm you own the site or have permission to test it.
• Hard limits: 500 requests/second, 100,000 requests and 1 hour per test.
• Request timeout and redirects are controlled, and one request is made per scheduled tick.
• Test settings reach k6 through environment variables, never by building a shell command, so a URL can't be turned into code.

Get started in 3 minutes

1. Install k6 (once)

RequestPulse uses k6, a free, open-source load-testing engine from Grafana Labs, to generate the traffic. You don't write any k6 code.

Restart VS Code afterwards so it picks up the new PATH. Installed k6 somewhere unusual (for example, you unzipped it)? Set the full path to k6.exe in the requestpulse.k6Path setting.

If k6 is missing, RequestPulse tells you and shows exactly what to do:

2. Open the dashboard

Press Ctrl+Shift+P (Cmd+Shift+P on macOS) and run:

RequestPulse: Open Dashboard

The badge at the top right should say k6 ready.

3. Run your first test

1. Start your app, or have a deployed URL ready.
2. Enter the URL, for example http://localhost:3000/api/products.
3. Choose a modest rate (10 requests per second for 30 seconds is plenty to begin).
4. Click Start test.

Start small. You can always go bigger.

What can I use it for?

Reading the results

Requests are not users. One visitor loading a page can trigger dozens of requests, so "20 requests per second" is not "20 users".

Commands and settings

Be honest about what a test can tell you

A load test is only as meaningful as its conditions. RequestPulse keeps these limits visible, and you should too:

• Your computer and network are part of the test. Requests start from your machine, so results include the path from you to the server. If the app and k6 run on the same PC, they compete for CPU.
• One test is not "maximum capacity". It shows how an endpoint behaved under one traffic pattern. To find a limit, run several tests at increasing rates and see where your targets stop being met.
• A URL does not reveal server CPU or memory. RequestPulse sees responses, not what is happening inside the server.
• Slow does not mean "the database". The report says what happened, not why. Use your server logs for the same time window to find the cause.
• Free hosting has quotas. Vercel, Render and similar plans have usage limits, and Render's free services sleep when idle, so the first requests of a test can be slow.
• Test traffic is real traffic. If your site counts page views on the server, a test will count too. See the FAQ.

Responsible use

Only test systems you own or have explicit permission to test. Load tests add real traffic to live services, can use up quotas, and may breach a host's terms of service. RequestPulse is built for testing your own applications. It is not a tool for attacking anyone else's, and it won't ask you to hide who is sending the traffic.

Privacy

• Nothing leaves your computer. There is no telemetry, no analytics and no account. RequestPulse contacts only the URL you choose to test.
• Results are saved as plain JSON files in the extension's local storage folder. Open it any time with RequestPulse: Open Saved Reports Folder.
• RequestPulse turns off k6's own usage reporting.
• URLs containing a username or password are rejected, so credentials are never saved to disk.

FAQ

Do I need to know k6? No. RequestPulse writes and runs the test for you. You only need k6 installed.

Does it work with Vercel and Render? Yes, with any publicly reachable http:// or https:// URL. You'll be asked to confirm you have permission to test it.

Will a test count towards my site's visitor counter? It depends on how the counter works. RequestPulse sends plain HTTP requests and does not run JavaScript. A counter that increments on the server when a URL is requested (middleware, a route handler, a GET endpoint) will count test requests. A counter that increments from browser JavaScript, or from third-party analytics, won't. Static pages served from a CDN never reach your code, so they won't count either. If you don't want test traffic counted, test against localhost or a staging copy.

Why only GET and HEAD? This version tests read endpoints. Logins, forms and multi-step journeys are on the wishlist below.

Why can't I send more than 500 requests per second? A single laptop or home connection can't reliably produce more, and results above that would mostly measure your own machine. The limits keep tests honest and protect other people's services.

The report says my target "was not fully delivered". What does that mean? k6 could not send requests as fast as you asked, because it ran out of virtual users (usually because responses were slow and requests piled up) or because this computer or network was the bottleneck. Treat the response times with care: the server saw less traffic than you requested.

Can I test an app that needs a login? Not yet. Version 1 tests URLs that don't need authentication.

Troubleshooting

Still stuck? Open an issue on GitHub and include the message you see.

Roadmap

Ideas on the list. These are possibilities, not promises:

• Gradually increasing traffic, to find where your targets stop being met
• Multi-step user journeys (log in, browse, add to cart)
• Live-updating charts while the test runs
• POST requests and custom headers
• Exporting a report as HTML or CSV
• Scheduled tests

Want one of these sooner? Tell me by opening an issue.

Platforms

Automated tests run on Linux with k6 1.6. The dashboard and k6 detection have also been checked on Windows 11. macOS uses the same code and has not been tested yet. If something misbehaves on your system, please report it.

Contributing

Issues and pull requests are welcome.

git clone https://github.com/parth-0702/requestpulse.git
cd requestpulse
npm install
# Press F5 in VS Code to launch an Extension Development Host
npm test          # unit tests; end-to-end tests run automatically when k6 is installed

Credits and licence

RequestPulse is released under the MIT licence.

Traffic is generated by k6, an open-source project by Grafana Labs. k6 is a trademark of Grafana Labs. RequestPulse is an independent project and is not affiliated with or endorsed by Grafana Labs or Microsoft.

If RequestPulse helped you catch a problem before your users did, a review on the Marketplace goes a long way.