Potpie allows you to chat with your codebase or any public codebase for that matter, to explain it, debug it, or do engineering tasks on it like generating test plans.

Potpie's Agents understand your codebase by breaking down your code into its constituent parts and building a knowledge graph from your code’s components. These specialized agents are:

Codebase Q&A Agent
Answers questions about your codebase using the knowledge graph and code analysis tools, from specific functions to overall architecture.

Unit Test Agent
Generates test plans and unit tests for specific functions or classes in your codebase.

Integration Test Agent
Generates integration test plans and test code for code snippets from the knowledge graph.

Debugging with Knowledge Graph Agent
Analyzes stacktraces and errors to provide debugging directions, helping you to iteratively find the root cause of issues.

Code Changes Agent
Generates detailed analysis of code changes in your current branch compared to the default branch

Code Generation Agent
An agent specialized in generating code for new features or fixing bugs.

Low-Level Design Agent
An agent specialized in generating a low-level design plan for implementing a new feature.

From the sound of it, it really looks like a super tool, the ultimate programmer's side kick. But is it so?

I've put it to the test to find out. First let's talk about how to get access to it. Since it is an open source tool, you can download the code from its Github repo and self-host it. Then there's the cloud version which requires no setup and you can just log into it using your Github credentials. 50 Free credits are provided for you to try it out.

After login in, you are presented with a main menu of the available agents and a input box which expects you to enter the repository to work on.

I tried out the Codebase Q&A agent on the Spring Boot Pet Clinic, a very popular application, to better understand it. In particular, I asked the Agents to "explain the relationships between the JPA entities". This is what it produced:

Spot on.

Then I instructed it to write a unit test for VetController.java, which it answered with :

Here's a unit test for the showVetList method in VetController.java using JUnit and Mockito:

Explanation:

• MockMvc: Used to perform HTTP requests and verify responses.
• Mockito: Mocks the VetRepository to return a paginated list of vets.
• Test Setup: Initializes the mocks and injects them into the VetController.
• Test Method: Mocks a GET request to /vets.html and verifies the status, view name, and model attributes.

This test ensures that the showVetList method correctly interacts with the VetRepository and returns the expected view and model attributes.

Nice. There's also an option to write your own Custom tests. Sample use cases include:

• Analyzing code to suggest optimizations and improvements
• Automatically generating boilerplate code for new features
• Conducting code reviews and providing feedback on best practices
• Extracting and summarizing documentation from code comments
• Identifying and reporting security vulnerabilities in the codebase
• Refactoring code to enhance readability and maintainability
• Creating unit tests based on existing code logic
• Assisting in migration of code to newer frameworks or libraries

For Potpie to fuel its agents it has multi-LLM support: OpenAI, Gemini, Anthropic, Deepseek, and Llama — simultaneously. 

The Github repo has already amassed 2.8K stars. Will yours be the next one?

More Information

Potpie

Potpie Github

Related Articles

HuggingFace AI Agents Course

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info