Reproducible AI model auditing, self-hosted

Name: SimpleAudit Studio
Author: SimpleAudit Studio

Freeze every audit input — scenarios, model config, judge profile — so results stay verifiable years later. Built on the Target → Auditor → Judge engine.

Get started View on GitHub

Open source · MIT licensed

Frozen inputs

Every run pins exact scenario content, model config, and judge profile. Re-run any audit, get identical results.

Any model endpoint

Connect Ollama, vLLM, OpenAI, Together, Groq, or any OpenAI-compatible API. Credentials stay out of the database.

Compare & export

Put audits side by side across models and prompts. Export CSV or JSON for downstream analysis.