Metadata-Version: 2.4
Name: unclejean
Version: 0.0.1
Summary: Client SDK for UncleJean.ai — continuous evaluation and control of LLM answers in production. Placeholder release: the name is reserved while the product is built.
Project-URL: Homepage, https://unclejean.ai
Project-URL: Documentation, https://unclejean.ai
Author-email: Jean-Christophe Capelli <jean@unclejean.ai>
License: MIT
Keywords: evaluation,guardrails,hallucination,llm,llmops,observability
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: System :: Monitoring
Requires-Python: >=3.9
Description-Content-Type: text/markdown

# unclejean

Client SDK for [UncleJean.ai](https://unclejean.ai) — continuous evaluation and control of
the answers your LLM gives in production: hallucination, sycophancy, toxicity and drift,
measured off the request path, with a circuit breaker on the answer.

## Status

**This is a placeholder release. There is no working client yet.**
The name is reserved while the product is built. The first design-partner integration is
targeted for autumn 2026.

Installing this package today gives you nothing but this notice. That is deliberate: we would
rather reserve the name honestly than ship a stub that pretends to work.

If you want to follow the work, or talk to us before there is anything to install:
<https://unclejean.ai>

## Intended interface

```python
from unclejean import UncleJean

client = UncleJean(OpenAI())   # one line to change, nothing else
```

Analysis runs off the request path: your users never wait for us, and our outage can never
become yours.

---

UncleJean.ai — Chevreuse, France.
