Closed
Description
Initial Checks
- I confirm that I'm using the latest version of Pydantic AI
- I confirm that I searched for my issue in https://github.com/pydantic/pydantic-ai/issues before opening this issue
Description
When using BedrockConverseModel with an Agent, if no system prompt is provided, the Bedrock API call fails because it sends the system key with an empty value, which Bedrock does not support. While prompts are a fundamental part of agents, this should not cause a failure, or the system_prompt in the Agent model should be made required.
Bedrock Payload
{
"modelId":"us.amazon.nova-pro-v1:0",
"messages":[
{
"role":"user",
"content":[
{
"text":"What is the capital of Italy?"
}
]
}
],
"system":[
{
"text":""
}
],
"inferenceConfig":{
}
}Error
botocore.exceptions.ParamValidationError: Parameter validation failed:
Invalid length for parameter system[0].text, value: 0, valid min length: 1
Example Code
from pydantic_ai import Agent
from pydantic_ai.models.bedrock import BedrockConverseModel
model = BedrockConverseModel(model_name="us.amazon.nova-pro-v1:0")
agent = Agent(model=model)
result_sync = agent.run_sync('What is the capital of Italy?')
print(result_sync.data)Python, Pydantic AI & LLM client version
(venv) ➜ python --version
Python 3.12.8
(venv) ➜ pip freeze |grep pydantic
pydantic==2.11.2
pydantic-ai==0.0.53
pydantic-ai-slim==0.0.53
pydantic-evals==0.0.53
pydantic-graph==0.0.53
pydantic-settings==2.8.1
pydantic_core==2.33.1