Add Nyraxis as a custom component in your Haystack pipeline to guard LLM inputs.

Setup

pip install haystack-ai nyraxis

export NYRAXIS_API_KEY="your-api-key"
export NYRAXIS_PROJECT_ID="your-project-id"

Code Example

from haystack import Pipeline, component, Document
from haystack.components.generators import OpenAIGenerator
from nyraxis import Nyraxis

@component
class NyraxisGuard:
    def __init__(self):
        self.nx = Nyraxis()

    @component.output_types(prompt=str)
    def run(self, prompt: str):
        result = self.nx.evaluate(prompt)
        if result.blocked:
            raise ValueError(f"Blocked by Nyraxis: {result.reason}")
        return {"prompt": prompt}

pipe = Pipeline()
pipe.add_component("guard", NyraxisGuard())
pipe.add_component("llm", OpenAIGenerator(model="gpt-4"))

pipe.connect("guard.prompt", "llm.prompt")

response = pipe.run({"guard": {"prompt": "Explain quantum computing"}})
print(response["llm"]["replies"][0])

What Gets Protected

Every prompt is evaluated by Nyraxis before reaching the LLM component
Blocked prompts raise an exception, preventing any downstream processing
Works with any Haystack generator (OpenAI, Hugging Face, custom)
Can be placed at multiple points in complex RAG pipelines

Haystack

Setup

Code Example

What Gets Protected

On this page