Skip to main content
agent-qa on GitHub,904stars
QA for the future
Changelog

Introducing agent-qa

Vostride Team

We’re introducing agent-qa: a QA harness for developers and coding agents, built to test real web and mobile applications.

Describe a journey in plain language, run it against your app, and inspect the actions, screenshots, and checks behind the result.

Why we built it

Coding agents are making software faster to produce. Knowing whether it works still takes time and attention. We built agent-qa to help close that gap: a harness that can exercise real applications and bring evidence back to the people responsible for shipping them. It is our first step toward the verification layer described in our manifesto, where humans and agents work together to ship quality confidently.

Read our manifesto ↗

Inside the dashboard

The local dashboard brings the testing workflow into one place. Browse run history, write and debug tests, and inspect the evidence behind each result. The tour below also shows suites, memory, insights, and workspace configuration.

See recent runs, filter by status or target, and open the execution you want to inspect.
agent-qa Runs table with search, status filters, queue slots, targets, durations, and recent test results
Review completed steps, captured browser state, and the action details behind a result.
A completed agent-qa run with passed steps, a captured browser screenshot, and per-action details
Follow an active run through its progress, action attempts, and verifier feedback.
agent-qa live execution view showing run progress, a step timeline, and verifier feedback
Define the target, setup, and ordered steps, then validate and run your test.
agent-qa test builder with target selection, setup hooks, and ordered natural-language test steps
Keep your test, browser, and network activity side by side while developing a journey.
agent-qa live test editor with completed steps, a connected browser, and network activity side by side
Group related tests into a repeatable suite with a shared target, setup, and teardown.
agent-qa suite builder showing a shared target, hooks, included tests, and run controls
Inspect what the agent has learned about your product, suites, and tests.
agent-qa Memory view with product observations, suite and test sections, confidence filters, and date filters
Review run reliability, duration, token usage, and memory activity across your workspace.
agent-qa Insights dashboard with pass rate, run duration, token usage, and memory activity charts
Set shared execution defaults and manage the workspace settings behind your tests.
agent-qa Configuration page with default model selection, parallel execution, and runtime settings
Move between runs, tests, hooks, suites, memory, insights, and configuration from one workspace.
agent-qa dashboard sidebar linking Runs, Tests, Hooks, Suites, Memory, Insights, and Config beside the Runs view

We’re just getting started

This first release lays the foundation for the verification layer we want to build. We’ll keep shaping agent-qa around real applications, difficult test journeys, and the teams using it. Try it on a flow you care about, inspect the results, and tell us where it falls short. We want to build this with you.

Get startedExplore the dashboardShare feedback on GitHub
← All updates

Give your coding agent a QA agent.

Open the agent quickstartView on GitHub