Installation

Basic Installation

Install pydantic-ai-guardrails using pip:

pip

pip install pydantic-ai-guardrails

uv

uv add pydantic-ai-guardrails

poetry

poetry add pydantic-ai-guardrails

This installs the core library with all built-in guardrails.

Optional Dependencies

Additional features are available through optional dependencies:

Telemetry (Logfire/OpenTelemetry)

For observability and tracing:

pip install pydantic-ai-guardrails[telemetry]

This adds:

• opentelemetry-api
• opentelemetry-sdk

Pydantic Evals Integration

To use pydantic_evals evaluators as guardrails:

pip install pydantic-ai-guardrails[evals]

This adds:

• pydantic-evals

All Features

Install everything:

pip install pydantic-ai-guardrails[all]

Third-Party Integrations

For additional guardrail capabilities, you can install these complementary libraries:

llm-guard

Battle-tested ML models for prompt injection, toxicity, and sensitive data detection:

pip install llm-guard

See llm-guard integration for usage.

autoevals

LLM-powered evaluations from Braintrust:

pip install autoevals

See autoevals integration for usage.

Requirements

• Python: 3.10 or higher
• pydantic-ai: 1.39.0 or higher

Verify Installation

Check that everything is installed correctly:

import pydantic_ai_guardrails

print(pydantic_ai_guardrails.__version__)

You can also verify the available guardrails:

from pydantic_ai_guardrails.guardrails.input import (
    length_limit,
    pii_detector,
    prompt_injection,
)
from pydantic_ai_guardrails.guardrails.output import (
    secret_redaction,
    llm_judge,
)

print("Installation successful!")

Next Steps

Now that you have pydantic-ai-guardrails installed, head to the Quick Start guide to create your first guarded agent.