Potpie AI is an open-source platform designed to create custom, task-oriented AI agents for your codebase. These agents perform a variety of engineering tasks with high precision, leveraging intelligence and context derived from your data. Potpie supports use cases such as system design, debugging, integration testing, and onboarding.
The platform builds a comprehensive knowledge graph of your code, enabling its agents to understand complex relationships within the codebase. This deep understanding allows the agents to assist with debugging, feature development, and other engineering tasks.
Potpie offers a suite of prebuilt agents, including:
- Debugging Agent: Analyzes stack traces and provides debugging steps specific to your codebase.
- Codebase Q&A Agent: Answers questions and explains functions, features, and architecture.
- Code Changes Agent: Analyzes code changes, identifies affected APIs, and suggests improvements.
- Integration Test Agent: Generates integration test plans and code.
- Unit Test Agent: Creates unit test plans and code to enhance test coverage.
- LLD Agent: Provides low-level design for new features based on functional requirements.
- Code Generation Agent: Generates and refactors code, and suggests optimizations.
Potpie integrates seamlessly with your existing development workflow and supports codebases of any size or language. It also offers a VSCode extension for direct integration into your development environment, allowing you to access and interact with Potpie agents without leaving your editor.
Potpie's tooling system includes functions for retrieving code snippets, executing knowledge graph queries, detecting code changes, and more. Users can build custom agents by defining system instructions, agent information, tasks, and tools, which can then be accessed via an API key for integration into CI/CD workflows and automated processes.
To get started with Potpie, you need Docker, Git, and Python 3.10. The setup involves initializing the UI submodule, configuring the environment, building the frontend, and starting the application. Once set up, you can interact with agents through API endpoints, monitor parsing status, and manage conversations.
Potpie is designed to be flexible and customizable, allowing users to create personalized agents and integrate additional tools. Contributions to the project are welcomed and governed by the Apache 2.0 License.
