Skip to content

Length Limit

The length_limit guardrail prevents overly long prompts that could be expensive, slow, or abusive.

Import

from pydantic_ai_guardrails.guardrails.input import length_limit

Basic Usage

from pydantic_ai_guardrails import GuardedAgent
from pydantic_ai_guardrails.guardrails.input import length_limit

guarded_agent = GuardedAgent(
    agent,
    input_guardrails=[
        length_limit(max_chars=1000),
    ],
)

Parameters

Parameter	Type	Default	Description
`max_chars`	`int \| None`	`None`	Maximum character count
`max_tokens`	`int \| None`	`None`	Maximum token count (requires tiktoken)
`encoding`	`str`	`'cl100k_base'`	Tiktoken encoding for token counting

Examples

Character Limit

# Block prompts over 2000 characters
guardrail = length_limit(max_chars=2000)

Token Limit

# Block prompts over 500 tokens
# Requires: pip install tiktoken
guardrail = length_limit(max_tokens=500)

Both Limits

# Block if either limit is exceeded
guardrail = length_limit(
    max_chars=2000,
    max_tokens=500,
)

Custom Encoding

# Use GPT-4 encoding
guardrail = length_limit(
    max_tokens=500,
    encoding='cl100k_base',  # Default for GPT-4
)

Violation Result

When triggered, returns:

{
    'tripwire_triggered': True,
    'message': 'Input exceeds maximum length of 1000 characters',
    'severity': 'medium',
    'metadata': {
        'length': 1500,
        'max_length': 1000,
        'limit_type': 'chars',
    },
}

Use Cases

Cost control: Prevent expensive long prompts
Abuse prevention: Block prompt stuffing attacks
Performance: Ensure reasonable response times
Rate limiting: Indirect limit on resource usage