类class
agentscope.formatterMoonshotChatFormatter
class MoonshotChatFormatter(_OpenAIFormatterBase)
源码Source
agentscope/formatter/_moonshot_formatter.py:63–309Moonshot 聊天格式化器,在智能体、模型与前端之间传递结构化消息和事件。位于 agentscope.formatter 包,继承自 _OpenAIFormatterBase。核心成员包括 format。
Moonshot's API is OpenAI-compatible, but thinking models (kimi-k2.6,
kimi-k2-thinking) return a reasoning_content field alongside
content in assistant messages. This formatter preserves that field
when re-sending assistant messages back to the API so that the
Preserved Thinking feature works correctly in multi-turn conversations.
属性摘要Attribute Summary
| 属性Attribute | 类型Type | 说明Description | 值Value |
|---|---|---|---|
input_types | list[str] | 输入类型字段。 | Field(default_factory=lambda: ['text/plain', 'image/*', 'audio/*', 'video/*'], description='The supported input types. Defaults to ``["text/plain", "image/*", "audio/*", "video/*"]``.') |
方法摘要Method Summary
| 方法Method | 说明Description |
|---|---|
async def format(msgs: list[Msg]) -> list[dict[str, Any]]async | 设置格式。 Format messages into the Moonshot / OpenAI-compatible API format. |
属性详情Attribute Detail
input_types
list[str] = Field(default_factory=lambda: ['text/plain', 'image/*', 'audio/*', 'video/*'], description='The supported input types. Defaults to ``["text/plain", "image/*", "audio/*", "video/*"]``.')输入类型字段。声明为 list[str]。
The supported input types.
源码Source
agentscope/formatter/_moonshot_formatter.py:73方法详情Method Detail
format
async def format(msgs: list[Msg]) -> list[dict[str, Any]]
async
设置格式。
Behaves identically to :class:OpenAIChatFormatter except that :class:ThinkingBlock content is placed into the reasoning_content field of the assistant message dict (required for Preserved Thinking).
参数Parameters
| 名称Name | 类型Type | 说明Description |
|---|---|---|
msgs | list[Msg] | 消息列表(类型 `list[Msg]`)。The list of message objects to format. |
list[dict[str, Any]]
返回dict[str,任意]列表。`list[dict[str, Any]]`:
源码Source
agentscope/formatter/_moonshot_formatter.py:105–309