VegaLoad is a free, open-source load testing tool for agentic developers. Your agent writes the test, runs it, and gives you a report.
TL;DR
- VegaLoad is a free, open-source load testing tool released under the Apache 2.0 license, sponsored by Harness and built by the Harness Resilience Testing team.
- Your agent writes the load test, runs it with VegaLoad, and returns a pass or fail result and one HTML report.
- Tests are files that live next to your code, run in CI and CD, and stay on your machine.
VegaLoad is a free, open-source load testing tool. You ask your agent for a test. It writes the file, runs it, and gives you a report.
What is VegaLoad?
Today, Harness is launching VegaLoad. VegaLoad is an open-source load testing tool for developers who build with AI agents. It is released under the Apache 2.0 license. Harness sponsors it, and the Harness Resilience Testing team builds it.
Here is how it works. You ask your agent for a load test. The agent writes a test file next to your service code. It runs the test with VegaLoad. You get a pass or fail result and one HTML report.
What you get:
- One small binary. It runs on macOS, Linux, Windows, Docker and Kubernetes. There is no JVM and no Node runtime to install.
- 14 protocols. HTTP/1.1, HTTP/2, gRPC, WebSocket, MQTT, Kafka, PostgreSQL, MySQL, Redis, RabbitMQ, FTP, SFTP, TCP and UDP.
- Pass or fail gates. Set limits for p95 latency, error rate and checks. A broken limit stops the pull request in CI, or the rollout in CD.
- Safe by default. Your agent can test localhost freely. Any other host needs your approval. Every run goes to an audit log on your machine.
- Ready for agents. A built-in MCP server and agent skills work with Cursor and Claude Code. One command, vegaload init, sets them up.
- Private. Tests run on your machine. The report is a single HTML file that opens offline.
Why we built it
AI agents now write a large share of code. Code ships faster than ever. Load testing has not kept up.
In many teams, a load test lives in a separate tool. The scripts sit in another repo. To get a run, someone files a ticket and waits. So many services reach production with no load test at all.
We think a load test should be a file. It should sit next to the code it tests. It should be reviewed in the same pull request. It should run in CI before the merge, and in CD after the deploy. And your agent should be able to write it and run it, safely, without leaving the editor.
We chose open source on purpose. Developers should be able to try a load test in minutes, read the code, and change what they need. We have seen how well this works with LitmusChaos, the open-source chaos engineering project that our team helps maintain. VegaLoad takes the same open approach to load testing.
How VegaLoad fits with Harness Resilience Testing
VegaLoad is built to grow a community. We will build it in the open, with the people who use it. We will ship the fixes and features that come from real use.
We have done this before. LitmusChaos is open source and has a large community. VegaLoad follows the same model for load testing. Together, they give developers two free, open-source tools. LitmusChaos is for chaos. VegaLoad is for load. Both work well on their own.
Some teams will want to go further. They need load from many machines in a cluster. They need shared history across teams. They need chaos resilience probes, to see how a service behaves when things break. That is where Harness Resilience Testing comes in.
The choice is yours. VegaLoad and LitmusChaos are free, and many teams will never need more. If you do want to scale, Harness Resilience Testing is always there. We are building a smooth path, so your VegaLoad tests move over and you do not start over.
Get started
Install VegaLoad, then run one command in your project:
brew install vegaload/tap/vegaloadvegaload init
Then ask your agent for a load test. Learn more at vegaload.ai. The code is at github.com/vegaload/vegaload.
Also see the install guide and the companion post, LitmusChaos Founders Launch VegaLoad.
