Hercules is an open-source testing agent designed for automating UI, API, security, accessibility, and visual validations without requiring coding expertise. It targets teams seeking to streamline end-to-end testing, particularly for complex applications like Salesforce, by translating Gherkin specifications into automated tests.

How It Works

Hercules employs a multi-agent architecture built on the AutoGen framework, featuring a Planner Agent for task decomposition and a Browser Navigation Agent equipped with a library of sensing and action tools. It prioritizes using pre-defined, conversational tools over LLM-generated code for safer and more predictable execution. DOM Distillation is used to refine the HTML DOM, focusing on relevant elements for LLM processing, and leverages the Accessibility Tree for improved accuracy.

Quick Start & Requirements

• Installation: pip install testzeus-hercules followed by playwright install --with-deps.
• Prerequisites: Python 3.11+, Playwright. LLM API keys (e.g., OpenAI, Anthropic, Groq, Mistral) are required for agent functionality.
• Setup: Can be done via PyPI, Docker, or source. A helper script automates setup on Linux/macOS.
• Resources: Requires LLM API access. Detailed setup guides and video tutorials are available.

Highlighted Details

• No-code test automation using Gherkin.
• Supports UI, API, Security (via Nuclei integration), and Accessibility (WCAG compliant) testing.
• Features auto-healing capabilities for reduced maintenance.
• Includes mobile device emulation and advanced tool integration (e.g., geolocation).

Maintenance & Community

The project is community-driven with an active Slack channel for support and discussion. Contributions are welcomed, with plans for a bounty program.

Licensing & Compatibility

Licensed under AGPL v3, which is a strong copyleft license. This may impose restrictions on linking with closed-source software.

Limitations & Caveats

The project is AI-native and relies heavily on LLM performance, with potential costs associated with API usage. While it aims for no-code, complex configurations and LLM interactions might require technical understanding. Video recording is not supported on all remote browser platforms.