# Sentinel-AI

> Sits beside your LLM and scores every prompt/response pair 0 to 100, with the claims and flags that produced the score, and a policy that decides serve, fix or block.

- Status: LIVE
- Period: 2025 to 2026
- Stack: Python, FastAPI, embeddings, PyPI SDK, Next.js docs
- Live: https://sentinelaihq.com
- Source: https://github.com/Blacksujit/Sentinel-AI
- HTML: https://www.sujit.top/missions/sentinel

## Objective

Production LLM apps need to know, per response, whether to serve it, fix it or stop it (and *why*), without putting another model in the generation path.

## Baseline

Ship the model's output as-is, or wrap it in a generic moderation endpoint that returns a label with no reasons.

## System

Model-agnostic: Sentinel observes prompt/response pairs and never sits in the generation path. Three detectors per request:

| Detector | Side | Checks |
|---|---|---|
| Prompt anomaly | Prompt | Distribution shift vs. baseline prompts |
| Jailbreak RAG | Prompt | Embedding similarity to known jailbreak patterns |
| Output risk | Response | 8 categories: violence, hate, self-harm, illegal activity, misinformation, privacy, inappropriate, harmful instructions |

The SDK's `verify()` adds local **factual-claim checking**: it extracts numeric claims from the response, cross-checks each against the prompt context, marks it `consistent`, `unverified` or `contradicted`, and auto-corrects contradictions when the prompt holds the right value.

```python
result = client.verify(prompt="Our refund policy is 60 days.",
                       response="We offer 30-day refunds.")
result["status"]     # "hallucinated"
result["corrected"]  # "We offer 60-day refunds."
```

Scores map to policy: **0 to 24** serve · **25 to 59** auto-correct or review · **60 to 100** block and escalate. Blocking, monitoring, async-with-webhooks and self-hosted modes.

## Eval

_Data pending: Precision and recall per detector on a labelled set; latency added per verify() call._

## Anomalies

_Data pending: The most interesting thing that broke while building Sentinel: a false positive, a jailbreak that slipped through, a latency problem._

## Outcome

Live at [sentinelaihq.com](https://sentinelaihq.com), SDK on PyPI as [`sentinelai-risk`](https://pypi.org/project/sentinelai-risk/), docs on [GitHub Pages](https://blacksujit.github.io/Sentinel-AI/).