Aura : Agents + GIt + Loops OSS ADE
Open source IDE for controlling AI coding agents
Start new thread

Aura: Agents + Git + Intent Open Source - OSS IDE for controlling AI coding agents with built in loops

by•
Aura is not another chat box for coding. It is a Git-native IDE for working with AI coding agents. You can run agents, track their changes at the function and class level, compare the code against the original intent, and prove whether a task was actually completed before you commit. Git shows you lines changed. Aura shows you what changed in the logic.

Add a comment

Replies

Best

Congrats on the launch, Ashik. I’ve been using coding agents quite a bit, and honestly, reviewing what they did is often harder than asking them to do it. After a long run, I’m usually staring at a huge diff wondering whether the agent actually solved the task or just produced something that looks right. Being able to rewind one function instead of rolling back the whole session sounds useful.

I’m curious about one thing. If I rewind a function and other changes depend on it, will Aura show me that before I do it?

Maker

 
not in the confirm step yet. Rewind is AST-scoped, it reverts that one function to its prior snapshot without touching the code around it, and shows you a before/after preview, but it doesn't currently warn you "3 things call this" before you pull the trigger. The reverse-dependency graph that would power exactly that warning is already built (it drives the "this change affects X" cards elsewhere in Aura), so wiring it into the rewind pre-flight is the natural next step, and honestly your question is the nudge to prioritize it. I'd rather tell you what it does today than oversell it.

Comparing implementation against intent is much more useful than another diff viewer. How do you preserve intent as the task evolves? If the developer corrects an assumption halfway through, can Aura distinguish the superseded requirement from the current one, or does it treat the whole conversation as one source of truth?

Maker

 it treats the conversation as a sequence of timestamped statements, not one blob, and the current one is what gets matched, but today it tracks supersession by time and scope, not yet as an explicit link.

 noted thanks for the information

Maker

 Sure mate, Please let me know if you find any issues or create GitHub issues.

First
Previous
12