Can I talk to my codebase and hear the answer with live visuals?

Yes, on DiffGuardian Pro. In Analyze Project you hold a key, ask your repository a question, and hear the answer in plain English while a graph of the flow builds on screen, drawn from your real call graph. Typing works too, and every turn lands in the same history.

#How it works

1. DiffGuardian indexes your repository into a code graph on your machine the first time you open Analyze Project. Later sessions reuse it.
2. Hold the key shown in the voice dock, ask, and release.
3. You hear the answer, with the real files named as it goes.
4. A graph builds on the canvas beside the conversation. Drag nodes to rearrange them, click a node to trace its flow, and click empty space to clear.

Every answer names its sources: the repository's code graph, a knowledge base you attached, or an MCP server you connected.

#Why it is quick

You do not need to know where to look. Ask how authentication works, where a webhook is verified or what a service touches, and you get the explanation and the picture at the same time.

#Ask out loud or type

Prefer typing? The chat is there. Voice runs on your own OpenAI or ElevenLabs key, or your operating system's voice. See Voice and Understand your codebase quickly.

DiffGuardian is a desktop app for macOS, Windows and Linux.

#Limits

• Analyze Project is a Pro feature.
• Answers come from the index of your checkout. After pulling a large branch, let the index catch up before asking about the new code.
• Talking to a pull request is separate: it highlights the code it explains and does not draw a graph. See Talk to your pull request.