Examples

Real-world examples of using pydantic-ai-shields.

Security

Example	Description
Security Overview	Combining multiple shields for defense-in-depth
PII Detection	Detect personally identifiable information
Prompt Injection	Block injection and jailbreak attempts
Toxicity Check	Screen content for toxic language
JSON Schema Validation	Validate tool arguments

Cost Control

Example	Description
Cost Tracking	Token usage monitoring and budget enforcement

Quick Example

Python

from pydantic_ai import Agent
from pydantic_ai_shields import (
    CostTracking, PromptInjection, PiiDetector, SecretRedaction, NoRefusals,
)

agent = Agent(
    "openai:gpt-4.1",
    capabilities=[
        CostTracking(budget_usd=5.0),
        PromptInjection(sensitivity="high"),
        PiiDetector(),
        SecretRedaction(),
        NoRefusals(),
    ],
)

result = await agent.run("Hello!")

Each shield handles a specific concern, keeping your code clean and modular.