Exploratory Tester

A toolkit for automating exploratory testing of web applications.

Architecture

The project produces three Docker images, each serving a distinct purpose:

Image	Base	Purpose	Port
development	dockette/vagrant:debian-12	Full local development environment with all tools	22 (SSH)
builder	debian:12-slim	CI/CD build and test environment	22 (SSH)
service	debian:12-slim	Lightweight runtime for exploratory testing capabilities

The service container is the core runtime where the exploratory testing capabilities are installed. It uses Claude Code to drive browser-based exploratory testing via Playwright MCP. Infrastructure is defined as code via Packer HCL configurations that use the Ansible provisioner to configure each image through playbooks and reusable roles.

Project Structure

.
├── bin/
│   ├── create-image                  # Packer build wrapper
│   └── setup-image-requirements      # Base image setup (SSH, Python)
├── infrastructure/
│   ├── packer/
│   │   ├── development.pkr.hcl       # Development image definition
│   │   ├── builder.pkr.hcl           # Builder image definition
│   │   └── service.pkr.hcl           # Service image definition
│   ├── ansible/
│   │   ├── playbook-development.yml  # Development provisioning
│   │   ├── playbook-builder.yml      # Builder provisioning
│   │   ├── playbook-service.yml      # Service provisioning
│   │   └── roles/                    # Reusable Ansible roles
├── .github/workflows/                # CI/CD pipelines
├── Vagrantfile                       # Local development via Docker provider
├── build.sh                          # Build images
├── push.sh                           # Push images to Docker Hub
└── create-latest.sh                  # Create multi-arch manifests

Prerequisites

• Vagrant
• Ansible
• Colima
• Docker

Local Development

The Vagrant environment uses the Docker provider to run the development image locally:

vagrant up
vagrant ssh

Building

Images are built using the build.sh script, which runs Packer inside the builder container:

# Build for a specific architecture
./build.sh service arm64
./build.sh service amd64

# Build for all architectures
./build.sh service

# Push to Docker Hub
./push.sh service arm64
./push.sh service amd64

# Create and push multi-arch manifest
./create-latest.sh service

Replace service with development or builder to build the other images.

Usage

docker run --rm \
  -e TARGET_URL=https://example.com \
  -e ANTHROPIC_API_KEY=sk-... \
  -v $(pwd)/output:/output \
  svanosselaer/exploratory-tester-service:latest

Variable	Required	Description
TARGET_URL	Yes	URL of the web application to test
ANTHROPIC_API_KEY	Yes	Anthropic API key for Claude Code
TEST_PROMPT	No	Custom testing instructions (defaults to general exploratory testing)

Results and screenshots are written to the /output directory.