Potpie

Potpie is an open-source platform that creates AI agents specialized in your codebase, enabling automated code analysis, testing, and development tasks. It deeply understands your codebase by breaking down code into its constituting parts and building a knowledge graph out of your code's components. This allows the platform to generate inferences at every level of your codebase and comprehensively answer questions about it.

Potpie provides purpose-built agents that are experts on your codebase to perform engineering tasks. The platform offers both pre-built agents for common tasks and the ability to build custom agents using tools that interface with the knowledge graph. Key capabilities include debugging, code reviewing, code generation, onboarding, and automated testing.

Potpie offers several specialized agents including a Codebase Q&A Agent that answers questions about your codebase and explains functions, features, and architecture, a Debugging Agent that automatically analyzes stack traces and provides debugging steps specific to your codebase, a Unit Test Agent that automatically creates unit test plans and code for individual functions, an Integration Test Agent that generates integration test plans and code for flows to ensure components work together, a Code Changes Agent that analyzes code changes, identifies affected APIs and suggests improvements, an LLD (Low-Level Design) Agent that creates low-level designs for implementing new features, and a Code Generation Agent that generates code for new features and refactors existing code.

You can access Potpie through multiple interfaces: as a VSCode extension, via Slack integration, or through the API. To begin, you install the extension or bot, generate an API key for secure access, and parse your repository using the Parse API to obtain a project ID.

Potpie allows you to build AI agents for your codebase in minutes. However, the initial parsing step can take a few minutes for moderately sized projects as Potpie builds the knowledge graph.

Potpie integrates seamlessly into your development environment through multiple channels, including a VSCode Extension that brings agent capabilities directly into your editor, allowing questions and suggestions without context switching; a Slack Integration where you can add the Potpie bot to any Slack channel, tag it with questions, and receive responses directly in the thread; and API Access for programmatic custom integrations.

Yes, Potpie offers Slack integration. You can add the Potpie bot to any Slack channel, tag it with questions or prompts for your custom AI agents, and receive responses directly in the Slack thread. You can also provide live links in your prompts for the AI agent to reference through the Web Access feature.

Yes, Potpie provides a platform for you to build your own custom agents using tools that interface with the knowledge graph. You can automate highly specific, repetitive tasks in your workflow, such as creating new components with boilerplate code or scanning pull requests for missing error handling.

Potpie is flexible and handles codebases of any size or language.

Potpie uses a deep code understanding approach with a built-in knowledge graph that captures relationships between code components. This allows agents to understand complex relationships and assist with various tasks. The platform uses tools like get_code_from_node_id and ask_knowledge_graph_queries to precisely retrieve information from your codebase.

Common use cases include reverse-engineering functionality by having the Q&A agent explore relevant files, generating low-level designs for open issues, analyzing frontend codebases for responsiveness, and automating code review and testing workflows.