MangoFisher/TestBrain

What it solves

TestBrain addresses the manual effort and inconsistency in software testing by automating the generation, review, and analysis of test cases. It helps teams build a scalable AI assistant that can parse product requirements, analyze Java source code, and generate interface test cases using large language models.

How it works

Built on Django and LangChain, the platform uses a modular architecture where specialized AI agents handle specific testing tasks. It integrates with Milvus for vector storage, allowing the system to retrieve relevant project documentation, PRDs, or past test cases to provide context-aware results. Users can switch between different LLM providers (like DeepSeek or Qwen) for each agent, and the system includes built-in error handling to fix malformed JSON outputs from models.

Who it’s for

This tool is designed for software testing teams looking to integrate AI into their workflow to improve efficiency, standardize test case creation, and gain deeper insights from existing project documentation and source code.

Highlights

• Multi-Agent System: Includes dedicated agents for test case generation, review, PRD analysis, interface testing, and Java code analysis.
• Flexible LLM Integration: Supports multiple model providers with the ability to configure specific models per agent or switch them manually via the UI.
• Knowledge Retrieval: Uses Milvus and BGE-M3 embedding models to store and search through project documentation for better context.
• Extensible Architecture: Designed with a modular factory pattern, making it easy to add new models, agents, or custom processing workflows.