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.

The agent sandbox demo lets you deploy an agent repository into an isolated Arga sandbox with selected digital twins. The demo is intentionally separate from the main Validate UI while the workflow is being tested. Open it directly at /demo/agent-sandbox in the web app.

​
How it works

  1. Select a GitHub repository and branch.
  2. Arga automatically detects framework, deploy target, port, command, environment variable names, twins, and infra.
  3. The page shows Detecting settings while detection runs.
  4. Review and edit the detected settings.
  5. Deploy the agent and selected twins.
  6. Open the returned agent URL, twin URLs, or run dashboard.
If you edit a field, later detection results do not overwrite it. Click Reset to detected settings to reapply the latest detected values.

​
Repository requirements

For the first demo, your repository should include one of these:
  • Dockerfile
  • docker-compose.yml or docker-compose.yaml
  • .arga/sandbox.yml with an agent target
Example: 
agent:
  framework: langchain
  context_dir: .
  dockerfile_path: Dockerfile
  app_port: 8000
  app_command: uvicorn app.server:app --host 0.0.0.0 --port 8000
  invoke:
    path: /invoke
    method: POST
  env:
    MODEL_PROVIDER: anthropic
  twins:
    - slack
    - github
  infra:
    - postgres

​
Accessing a deployment

After deployment, Arga returns links for:
  • Agent surface: https://r<RUN_ID>--app.sandbox.argalabs.com
  • Twin surfaces: https://r<RUN_ID>--<twin>.sandbox.argalabs.com
  • Run dashboard: /runs/<RUN_ID>
The demo page also shows these links when the sandbox is ready.

​
Environment variables

Add runtime secrets as KEY=value lines in the demo. Arga passes them to the deployed agent container for that run. Run summaries show only environment variable names, not values.