Skip to content

CubeLoop — a Pythonic, async-native agent harness core designed for high performance, readability, and production-grade persistence.

feat(providers): add BoundModel.generate convenience method

xf gong745cd77039c8
diff --git a/cubepi/providers/base.py b/cubepi/providers/base.py
index 76af65e..58c7e48 100644
--- a/cubepi/providers/base.py
+++ b/cubepi/providers/base.py
@@ -95,6 +95,30 @@ class BoundModel:
provider: Provider
spec: Model
+ async def generate(
+ self,
+ messages: list[Message],
+ *,
+ system_prompt: str = "",
+ tools: list[ToolDefinition] | None = None,
+ options: StreamOptions | None = None,
+ max_output_tokens: int | None = None,
+ temperature: float | None = None,
+ thinking: ThinkingLevel | None = None,
+ thinking_budgets: ThinkingBudgets | None = None,
+ ) -> AssistantMessage:
+ return await self.provider.generate(
+ model=self.spec,
+ messages=messages,
+ system_prompt=system_prompt,
+ tools=tools,
+ options=options,
+ max_output_tokens=max_output_tokens,
+ temperature=temperature,
+ thinking=thinking,
+ thinking_budgets=thinking_budgets,
+ )
+
class TextContent(BaseModel):
type: Literal["text"] = "text"
diff --git a/tests/providers/test_bound_model_calls.py b/tests/providers/test_bound_model_calls.py
new file mode 100644
index 0000000..b600a37
--- /dev/null
+++ b/tests/providers/test_bound_model_calls.py
@@ -0,0 +1,82 @@
+from __future__ import annotations
+
+from typing import Any
+
+import pytest
+
+from cubepi.providers.base import (
+ AssistantMessage,
+ BaseProvider,
+ Message,
+ Model,
+ StreamOptions,
+ TextContent,
+ ToolDefinition,
+ UserMessage,
+)
+
+
+class _RecordingProvider(BaseProvider):
+ def __init__(self) -> None:
+ super().__init__(provider_id="rec")
+ self.generate_calls: list[dict[str, Any]] = []
+
+ async def generate( # type: ignore[override]
+ self,
+ model: Model,
+ messages: list[Message],
+ *,
+ system_prompt: str = "",
+ tools: list[ToolDefinition] | None = None,
+ options: StreamOptions | None = None,
+ max_output_tokens: int | None = None,
+ temperature: float | None = None,
+ thinking: Any = None,
+ thinking_budgets: Any = None,
+ ) -> AssistantMessage:
+ self.generate_calls.append(
+ {
+ "model": model,
+ "messages": messages,
+ "system_prompt": system_prompt,
+ "tools": tools,
+ "options": options,
+ "max_output_tokens": max_output_tokens,
+ "temperature": temperature,
+ "thinking": thinking,
+ "thinking_budgets": thinking_budgets,
+ }
+ )
+ return AssistantMessage(
+ content=[TextContent(text="ok")],
+ provider_id=model.provider_id,
+ model_id=model.id,
+ )
+
+
+async def test_bound_model_generate_forwards_to_provider() -> None:
+ provider = _RecordingProvider()
+ bound = provider.model("model-x", temperature=0.5)
+ messages = [UserMessage(content=[TextContent(text="hi")])]
+
+ response = await bound.generate(
+ messages=messages,
+ system_prompt="be brief",
+ max_output_tokens=64,
+ temperature=0.0,
+ thinking="off",
+ )
+
+ assert isinstance(response, AssistantMessage)
+ assert response.provider_id == "rec"
+ assert response.model_id == "model-x"
+
+ assert len(provider.generate_calls) == 1
+ call = provider.generate_calls[0]
+ assert call["model"] is bound.spec
+ assert call["messages"] is messages
+ assert call["system_prompt"] == "be brief"
+ assert call["max_output_tokens"] == 64
+ assert call["temperature"] == 0.0
+ assert call["thinking"] == "off"