Skip to main content

hackagent.core.contracts.completion

The result of one model call, and why a call failed.

RawExchange Objects​

class RawExchange(BaseModel)

The provider-level request and response, for callers that need them.

LLMError Objects​

class LLMError(BaseModel)

Why a call failed. Calls return errors as values.

GuardrailInfo Objects​

class GuardrailInfo(BaseModel)

A guardrail blocked the prompt (before) or the response (after).

Completion Objects​

class Completion(BaseModel)

The result of one model call.

Replaces the response dict: generated_text / processed_response become text; error_message / error_category become error; agent_specific_data usage, finish reason, model and invoked parameters become typed fields; raw_response_* and raw_request become raw; a guardrail envelope becomes guardrail.

ok​

@property
def ok() -> bool

True when the call produced a usable response.

blocked​

@property
def blocked() -> bool

True when a guardrail blocked or censored the call.

guardrail_info​

@property
def guardrail_info() -> Dict[str, Any]

Guardrail metadata as a plain dict (empty when not blocked).