← Back to Discover

Potpie

Open-source platform that maps an engineering organization's code, tickets and docs into a context graph and runs debugging, testing and impact-analysis agents on it inside the customer's own VPC.

Coding agents
Visit website ↗

What it does

Potpie positions itself as software-development-lifecycle automation for large engineering organizations. It connects sources such as GitHub, GitLab, Jira, Linear and Confluence, builds a graph of how services, files, tickets and decisions relate, and turns that into a typed model of the domain that is refreshed with every commit and ticket. Agents then work on that graph. The site shows root-cause analysis for a latency regression, impact analysis for a policy change and regression-test drafting, with results handed to a person for review. Answers are shown with links back to sources such as pull requests, tickets and documents. It is meant to run on the customer's own infrastructure, in a VPC and read-only by default, and to work with several models, including OpenAI and Gemini. The site says forward-deployed engineers embed with a team, test the setup there, then help replicate it across the organization. The code is open source and can be self-hosted.

Capabilities

  • Context graph linking code, tickets, docs and ownership
  • Code graph with syntax tree, call graph and data flow
  • Specialist agents for debugging, impact analysis, code generation and testing
  • Answers that link back to pull requests, tickets and documents
  • Runs in your own VPC, read-only by default
  • Works with several models and with IDEs and Claude Code

When to use it

  • A large engineering team needs root-cause analysis grounded in its own code and tickets
  • You want specs and regression tests drafted with full codebase context
  • You need to onboard engineers faster on a complex codebase
  • Your data must stay inside your own infrastructure