Potpie
Open-source platform for building custom AI agents on your codebase.
About
Potpie is an open-source platform that lets developers create AI agents specialized in their own codebase. It builds a comprehensive knowledge graph of the code — capturing relationships between components — which agents then query to perform debugging, testing, system design, onboarding, code review, and documentation tasks. It ships with prebuilt agents (including stack-trace analysis and integration-test generation), a chat interface, a VS Code extension, and multi-LLM support so teams can plug in their preferred models.
The honest caveat: as a younger open-source project, documentation and community support are thinner than established alternatives, and building an accurate knowledge graph for very large monorepos can be slow and resource-intensive. Custom agent quality depends heavily on prompt design skill.
What it can do
- Codebase knowledge graph for deep contextual understanding
- Prebuilt agents for debugging, testing, and code review
- Custom agents defined through simple prompts
- VS Code extension and chat interface
- Multi-LLM support for model flexibility
Setup
Clone from GitHub or install via the documented quickstart. Install the Potpie VS Code extension from the Marketplace for in-editor access. Configure your LLM API keys.
Prerequisites
Python environment. API keys for your chosen LLM provider. Sufficient disk and memory for knowledge-graph indexing on large repos.
Use cases
Works with
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 44 (tools-a); Trending score 62/100 from researcher assessment.
★ 2,800 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.