Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.argalabs.com/llms.txt

Use this file to discover all available pages before exploring further.

Get Arga running against your app in three steps. The CLI is the fastest way in — once you’re set up, connect your integrations so Arga can generate tests grounded in your real stack instead of just the prompt you give it.

​
Validate a URL

1

Install the CLI

uv tool install arga-cli
You can authenticate in two ways:
  • Email signup (quickstart): Sign up at app.argalabs.com/get-started with your email to get a quickstart API key. This key is good for 5 twin provisions — enough to try Arga out. The wizard will prompt you for the key.
  • Full login: Run arga login to authenticate via GitHub. This gives you a full-access key with no provision limit.
2

Run your first validation

Point Arga at any deployed URL and tell it what to test:
arga test-runner runs url \
  --url https://your-app.com \
  --prompt "test ..."
Arga spins up a browser, follows your prompt, and reports back with step-by-step results including screenshots.
3

Check your results

The CLI prints a run ID when the validation starts. Track progress with:
arga runs status <run_id>
If a run stalls or fails, inspect the current logs snapshot with:
arga runs logs <run_id> --errors-only
Or open app.argalabs.com to see results in the dashboard with screenshots and step details.

​
Testing flows that require sign-in

If the flow you want to test involves signing up, logging in, or any authenticated area of your app, pass the credentials Arga should use. Arga’s browser will use these to sign in before executing the test: 
arga test-runner runs url \
  --url https://your-app.com \
  --prompt "sign up for a new account and complete onboarding" \
  --email testuser@example.com \
  --password your-test-password
Both --email and --password must be provided together. These are the credentials for your app (not your Arga account) — use a test account you’ve set up on your staging environment, or provide new credentials if you’re testing the signup flow itself.
When testing signup, create credentials that don’t already exist in your app so Arga can register a fresh account. When testing login or authenticated flows, use credentials for an account that already exists.

​
Go further: deterministic sandboxes with API twins

Once you’ve run a basic URL validation, you can spin up a fresh sandbox with API twins: stateful replicas of Stripe, Slack, GitHub, Notion, and more that reset between runs so your tests are fully deterministic.
1

Provision twins for your project

From your project directory, run the wizard using the CLI or npx:
arga wizard init

npx arga-wizard
arga wizard init requires CLI v0.1.5+. If you’re on an earlier version, use npx arga-wizard instead.
arga wizard init automatically passes your saved API key if you’ve run arga login. Otherwise, the wizard prompts you for a key — paste the quickstart key from email signup or any API key from your settings.
Quickstart keys are limited to 5 twin provisions. Run arga login for full access when you’re ready.
The wizard will:
  1. Ask which services your app integrates with (Stripe, Slack, Discord, etc.)
  2. Back up your .env to .env.arga-backup and rewrite API keys to point at twins
  3. Provision ephemeral twin instances (typically under a minute)
Start your app normally — API calls to selected services now route through twins. Sessions last 10 minutes by default. Restore your original environment with cp .env.arga-backup .env.See the twins quickstart for details on available twins and session management.
2

Run a browser test

Open app.argalabs.com/demo_runner or choose Runs from the Test Runner section:
  1. Enter the URL where your app is running — this should be the staging or local URL that the wizard already configured to talk to twins (your .env was rewritten in step 1)
  2. Optionally describe what to focus on
  3. Arga opens a live browser, plans the flow, and generates editable blocks
  4. Watch the browser frame and event stream, then rerun or save the result as a reusable test
See Validate modes for the full Previews and Test Runner layout.

​
Testing AI agents?

If your team is building AI agents and you want to validate their behaviour — including prompt injection resistance, tool-use boundaries, and graceful degradation — Arga can test agents in sandboxed environments backed by digital twins. Agent testing is currently in private beta. See Agent & security testing for details on what Arga covers and why agent behaviour requires a different testing approach.

Discuss agent testing

Book a 30-minute session to discuss agent testing for your use case.

​
Next steps

Validate modes

Learn when to use Runs, Tests, Scenarios, PR Checks, Sandboxes, and Twins.

CLI

Full CLI reference for terminal validation workflows.

MCP

Set up Arga context and validation tools in your IDE agent.

How Arga works

Understand the architecture behind Arga’s testing pipeline.

Connect integrations

Connect GitHub, Slack, Jira, Sentry, Notion, and other context sources.