Skip to content

About

A production-quality, self-hostable platform for running reproducible AI model audits using the SimpleAudit engine (Target → Auditor → Judge).

Topics

Resources

Stars

2 stars

Watchers

0 watching

Forks

Latest commit

 

History

668 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

title SimpleAudit Studio
emoji 🔍
colorFrom blue
colorTo indigo
sdk docker
app_port 7860
pinned false

SimpleAudit Studio

License: MIT Python PyPI

A self-hostable platform for reproducible AI model audits using the SimpleAudit engine.

Quick run

Needs uv:

uvx simpleaudit-studio@latest

To run the latest code from GitHub instead of the published package:

uvx --from "git+https://github.com/SimulaMet/SimpleAuditStudio@main" spin

Open http://localhost:8000. Try the live demo if you do not want to install anything.

Development

git clone https://github.com/SimulaMet/SimpleAuditStudio
cd SimpleAuditStudio
cp .env.dev.example .env
uv sync --extra dev
uv run manage.py setup_local
uv run manage.py dev

For an existing instance, dev reuses it rather than restarting the entire stack. Run uv run manage.py dev status to see the actual port. Run uv run manage.py dev agent-info for detailed agent-safe instructions. dev down stops without data loss, dev reset recreates clean local data, and dev uninstall removes owned state and app-scoped uv caches. The same lifecycle actions are available as uvx --from simpleaudit-studio spin ACTION. See development lifecycle for isolation and safety details.

The app opens at http://localhost:8000 with live reload. Use --no-browser to skip opening a browser.

What it does

  • Build versioned scenario sets and register OpenAI-compatible models.
  • Define judges and run audits across models, scenarios, and settings.
  • Review runs live and compare results in the dashboard.
  • Monitor models for drift on a schedule.

Documentation

  • Deployment — all run commands, settings, chat, upgrades, backups, and production notes
  • Chat — chat setup and privacy details
  • Architecture — system design and service boundaries
  • Domain model — core data and invariants

Contributing

Architecture decisions are documented in docs/architecture.md.

About

A production-quality, self-hostable platform for running reproducible AI model audits using the SimpleAudit engine (Target → Auditor → Judge).

Topics

Resources

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages