Skip to main content

Tool Validation

Schema validation and type safety for tool inputs and outputs.

Overview​

Tool validation ensures that tool inputs and outputs conform to expected schemas using Pydantic and Instructor.

Basic Validation​

from recoagent.tools import ToolValidator, ValidationError
from pydantic import BaseModel

class SearchInput(BaseModel):
    query: str
    limit: int = 10

validator = ToolValidator()

# Validate input
try:
    validated_input = validator.validate_input(
        tool_name="search_tool",
        input_data={"query": "test", "limit": 5},
        schema=SearchInput
    )
except ValidationError as e:
    print(f"Validation failed: {e}")

Structured Validation with Instructor​

from recoagent.tools import StructuredToolValidator

validator = StructuredToolValidator()

# Validate with automatic retries
result = validator.validate_structured_output(
    tool_output=raw_output,
    schema=OutputSchema,
    max_retries=3
)

Best Practices​

  1. Define Clear Schemas: Use Pydantic for type safety
  2. Validate Inputs: Always validate before execution
  3. Handle Errors: Implement proper error handling
  4. Use Retries: Configure retries for validation failures