# NemotronPolicyNudgeMiddleware

> **Class** in `deepagents`

📖 [View in docs](https://reference.langchain.com/python/deepagents/profiles/harness/_nvidia_nemotron_3_ultra/NemotronPolicyNudgeMiddleware)

Inject lightweight policy nudges for common Ultra3 agent-control misses.

## Signature

```python
NemotronPolicyNudgeMiddleware()
```

## Extends

- `AgentMiddleware`

## Properties

- `name`
- `state_schema`

## Methods

- [`wrap_model_call()`](https://reference.langchain.com/python/deepagents/profiles/harness/_nvidia_nemotron_3_ultra/NemotronPolicyNudgeMiddleware/wrap_model_call)
- [`awrap_model_call()`](https://reference.langchain.com/python/deepagents/profiles/harness/_nvidia_nemotron_3_ultra/NemotronPolicyNudgeMiddleware/awrap_model_call)
- [`before_model()`](https://reference.langchain.com/python/deepagents/profiles/harness/_nvidia_nemotron_3_ultra/NemotronPolicyNudgeMiddleware/before_model)
- [`abefore_model()`](https://reference.langchain.com/python/deepagents/profiles/harness/_nvidia_nemotron_3_ultra/NemotronPolicyNudgeMiddleware/abefore_model)

---

[View source on GitHub](https://github.com/langchain-ai/deepagents/blob/90d4ce36dd6f9f402b8c40547972ffc13c8814ca/libs/deepagents/deepagents/profiles/harness/_nvidia_nemotron_3_ultra.py#L1100)