DiffGraph

What Inspired Me 💡

Code has always been one of the most challenging skills for me to master. As a visual learner myself, I struggled with the sheer mental heavy-lifting that standard code reviews demand.

I believe I speak for many in the developer community when I say that reviewing multi-file Pull Requests can be downright exhausting. You open a PR spanning ten different files, and suddenly you're forced to mentally trace complex function calls, API payloads, and database schema updates.

Reading a line-by-line text diff tells you what changed, but it leaves you to map out how the system actually behaves in your head (or on a scrap paper desk doodle that should’ve hit the recycle bin days ago).

I wanted to stop tracing code in my head and start visualizing it. That’s why I built DiffGraph.

The idea for DiffGraph came from asking a simple question: What if every Pull Request automatically generated an interactive, color-coded execution map directly inside the GitHub comment thread? Instead of having to memorize complex dependency chains, DiffGraph translates raw git diffs into live system architecture diagrams which highlights contract mismatches in red and successful execution paths in green.

It tells you what you’re doing wrong and gives you the visual satisfaction of what you’re doing right

What it does

DiffGraph is an automated developer bot that transforms dense, multi-file Pull Requests into clear, color-coded execution flowcharts directly inside your GitHub PR comments. Instead of forcing code reviewers to scroll through hundreds of lines of raw text, DiffGraph analyzes system changes in real time and generates a visual architectural diagram of your code's data flow.

Key Features & Capabilities

• Automated PR Flowcharts: Converts raw ⁠git diff⁠ code changes into interactive Mermaid.js diagrams embedded natively in GitHub comments without requiring external dashboards.
• Red-to-Green Error Detection: Automatically parses API contracts, payload models, and function calls across files. It highlights broken integrations or missing payload fields in Red and updates them to Green as soon as a fix is pushed.
• Multi-Layer Architecture Mapping: Visualizes how data moves across your entire stack, tracing calls from frontend UI components to backend API endpoints and database schemas.
• 100% Privacy-First & Zero Cost: Runs entirely on local infrastructure using Ollama and Llama 3.2, ensuring your private codebase never leaves local RAM and avoiding cloud API costs.
• Non-Blocking GitHub Webhooks: Utilizes asynchronous background processing to deliver fast webhook responses without timing out GitHub's connection

Built With

• data-visualization
• github
• graph-theory
• ollama
• privacy-first
• pygithub
• python
• software-architecture
• webhooks