Skip to main content

Client Integration Guide

Making Your Test Suite Agnostic-Ready​

This guide explains how to prepare any containerized automation suite (Python, Java, Node.js, etc.) so it can run safely and correctly inside the Agnox.

The key principle: The platform controls execution - your repo provides behavior.

The fastest way to get started is with the official Agnox CLI:

npx @agnox/agnox-cli@latest init

This interactive tool will:

  • Generate a Dockerfile, entrypoint.sh, and .dockerignore tailored to your framework
  • Auto-detect your Playwright version from package.json and pin the correct Docker base image
  • Build a multi-platform Docker image (linux/amd64 + linux/arm64) and push it to Docker Hub

Supported frameworks: Playwright (TypeScript/Node.js) and Pytest (Python).

📦 CLI Repository: github.com/agnox/agnox-cli ·· npm: @agnox/agnox-cli

2. Manual Setup (Advanced)​

If you prefer to set up your project manually, follow the steps below.

2.1 Mandatory entrypoint.sh​

For security and consistency, the Worker does not execute arbitrary commands. Instead, it always runs:

/app/entrypoint.sh <folder>

Your responsibility​

Create an executable entrypoint.sh at the root of your repo. This script acts as the bridge between the platform and your specific test runner.

Recommended Script Pattern:

#!/bin/sh

FOLDER=$1

# Remove local .env to enforce Worker configuration
if [ -f .env ]; then
  echo "Removing local .env to enforce injected configuration..."
  rm .env
fi

echo "Running against BASE_URL: $BASE_URL"

if [ -z "$FOLDER" ] || [ "$FOLDER" = "all" ]; then
  echo "Running ALL tests..."
  npx playwright test
else
  echo "Running tests in folder: $FOLDER"
  npx playwright test "$FOLDER"
fi

# Generate Allure HTML report from raw results
# NOTE: The Agnostic Worker handles report generation automatically!
# You do NOT need 'allure generate' here.

⚠️ Do not use exec before your test runner command. Using exec replaces the shell process and prevents the worker from correctly tracking the process exit.

Why this matters​

  • Security: Prevents configuration conflicts.
  • Predictability: Guarantees the test runs exactly as the Dashboard intended.
  • Flexibility: Allows folder-level test selection from the UI.

2.2 Dockerfile Requirements​

Your test suite must be containerized and published to a registry (Docker Hub, GHCR).

FROM mcr.microsoft.com/playwright:v1.50.0-jammy

WORKDIR /app

COPY package*.json ./
RUN npm ci

COPY . .

# Ensure entrypoint is executable
RUN chmod +x /app/entrypoint.sh

# Do NOT use ENTRYPOINT or CMD here.
# The Worker Service will inject the entrypoint command at runtime.

2.3 Environment Variables & Validation​

The platform injects environment variables only if they are whitelisted in the infrastructure.

Best Practice​

If you use validation libraries like Zod, ensure your schema allows for optional defaults or that you have added the variable to the infrastructure's INJECT_ENV_VARS list.

3. Register Your Project in the Dashboard​

Once your Docker image is built and pushed to a registry:

  1. Open the AAC Dashboard
  2. Go to Settings → Run Settings
  3. Create a new project and enter a project name
  4. Enter your Docker image name (e.g., youruser/my-automation:latest)
  5. Configure your environment URLs (Dev, Staging, Production)
  6. Set a default test folder (or leave as all)

These settings will pre-fill the Execution Modal each time you launch a new test run.

4. What You Should NOT Do ❌​

  • ❌ Run Playwright directly in Docker CMD.
  • ❌ Expect shell access to the server.
  • ❌ Read infrastructure-level secrets (like the VPS SSH key).
  • ❌ Depend on a local .env file inside the image.

5. What You CAN Do ✅​

  • ✅ Read injected environment variables (process.env.BASE_URL).
  • ✅ Control test selection via folders.
  • ✅ Use any framework (Playwright, Pytest, Robot Framework).

6. Using the Interactive Dashboard 🎮​

Once your image is integrated, you can utilize the Dashboard's advanced features:

Manual Execution (The Play Button)​

You don't need to trigger tests via API. You can launch them visually:

  1. Click the "Launch Execution" button (Top Right).
  2. Environment: Select Dev, Staging, or Prod - the system pre-fills the URL from your project settings.
  3. Folder: Type a folder path (e.g., tests/login) or select all.
  4. Launch: The test starts immediately, and you will see logs streaming in real-time.

🕵️ Troubleshooting with AI​

If a test fails, the system automatically performs a Root Cause Analysis.

  1. Look for the status: ANALYZING (Purple).
  2. Once finished, a ✨ Sparkle Icon will appear next to the FAILED status.
  3. Click the icon to open the AI Analysis Report.
    • See the exact error reason.
    • Get code snippets for suggested fixes.
    • Understand why it failed without reading 1000 log lines.

Client Integration Complete​

Your test suite is now fully agnostic, portable, and secure.