DeepKeep AI Firewall Guardrails

Use DeepKeep AI Firewall as custom guardrails for Agno Agents.

Use DeepKeep AI Firewall as custom guardrails for Agno Agents. DeepKeep runs before user input reaches the model and after model output is generated.

deepkeep_ai_firewall.py

"""
DeepKeep AI Firewall Guardrails
===============================

This example shows how to use DeepKeep AI Firewall as custom guardrails for
Agno Agents. DeepKeep runs before user input reaches the model and after model
output is generated.

Requirements:
- pip install agno-deepkeep
- DEEPKEEP_API_KEY and DEEPKEEP_BASE_URL set

Set credentials before running:

    export DEEPKEEP_API_KEY="dk_..."
    export DEEPKEEP_BASE_URL="https://api.example.deepkeep.ai"

Usage:
    python cookbook/02_agents/08_guardrails/deepkeep_ai_firewall.py
"""

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno_deepkeep import DeepKeepGuardrail

agent = Agent(
    name="DeepKeep Protected Agent",
    model=OpenAIResponses(id="gpt-5.2"),
    instructions="Answer user questions safely and concisely.",
    pre_hooks=[
        DeepKeepGuardrail(
            pre_model="input-firewall-id",
        )
    ],
    post_hooks=[
        DeepKeepGuardrail(
            post_model="output-firewall-id",
        )
    ],
    markdown=True,
)


if __name__ == "__main__":
    agent.print_response("Explain how to store API keys securely.")

Run the Example

1. Set up your virtual environment

   Mac

   uv venv --python 3.12
   source .venv/bin/activate
   
2. Install dependencies

   uv pip install -U agno agno-deepkeep openai
   
3. Export your OpenAI API key

   Mac/Linux

   export OPENAI_API_KEY="your_openai_api_key_here"
   
4. Run the example

   Save the code above as deepkeep_ai_firewall.py, then run:

   python deepkeep_ai_firewall.py
   

Full source: cookbook/02_agents/08_guardrails/deepkeep_ai_firewall.py