I built ARIA to address an emerging challenge in software development: defining clear boundaries for how AI assistants interact with our codebases.
As AI coding tools become more common, teams need a standardized way to specify which files AI can modify, which it can only analyze, and which require human review.
ARIA provides:
- YAML-based policies with inheritance support
- Pre-defined models (Guardian, Observer, Assistant)
- IDE integration for Windsurf and Cursor
- Conversion to IDE rules and ignore files
I built ARIA to address an emerging challenge in software development: defining clear boundaries for how AI assistants interact with our codebases.
As AI coding tools become more common, teams need a standardized way to specify which files AI can modify, which it can only analyze, and which require human review.
ARIA provides: - YAML-based policies with inheritance support - Pre-defined models (Guardian, Observer, Assistant) - IDE integration for Windsurf and Cursor - Conversion to IDE rules and ignore files
The project is early (alpha) but functional. Documentation is available at https://antenore.github.io/ARIA/
I'm curious if others are encountering similar governance challenges with AI in their workflows, and what approaches you've tried.