Aura – Semantic version control for AI-generated code

Summary: Aura provides semantic version control by hashing Abstract Syntax Trees (AST) instead of text lines, enabling precise tracking and management of AI-generated code. It offers tools to revert specific code changes, block undocumented commits, and optimize token usage, all running locally under an open-source Apache 2.0 license.

What it does

Aura parses codebases into ASTs for languages like Rust, Python, TypeScript, and JavaScript, enabling semantic diffs and operations. It includes features such as selective function rollback, AI chat memory reset, commit validation against code intent, execution planning, secure allowlisting, audit of undocumented changes, and token-efficient context compression.

Who it's for

It is designed for developers and teams using autonomous AI agents to generate code, particularly those facing challenges with traditional Git in managing non-linear, high-volume AI code changes.

Why it matters

Aura addresses the limitations of text-based version control for AI-generated code by preventing merge conflicts, ensuring traceability, and reducing costly token usage in AI workflows.