🔌 API 参考

本页记录 v0.3.1 当前 Python SDK 的实际 API 结构。

Client

Client 是主入口。

from wyseos.mate import Client, ClientOptions
from wyseos.mate.config import load_config

client = Client(load_config("mate.yaml"))
# 或
client = Client(ClientOptions(api_key="YOUR_API_KEY"))

ClientOptions

说明：

• base_url 必须以 http:// 或 https:// 开头。
• timeout 范围是 1-300。
• api_key 和 jwt_token 任选其一配置。

Client 上的服务

• client.user
• client.team
• client.agent
• client.session
• client.browser
• client.file_upload
• client.marketing
• client.product

Session API

通过 client.session 访问。

方法

• create(request: CreateSessionRequest) -> CreateSessionResponse
• get_info(session_id: str) -> SessionInfo
• get_messages(session_id: str, page_num: int = 1, page_size: int = 20, filter: Optional[MessageFilter] = None) -> GetMessagesResponse
• get_between_messages(session_id: str, from_message_id: str, to_message_id: str) -> GetMessagesResponse
• get_marketing_data(session_id: str, type: Optional[str] = None) -> Dict[str, Any]
• update_session_name(request: UpdateSessionNameRequest) -> None

CreateSessionRequest

from wyseos.mate.models import CreateSessionRequest

req = CreateSessionRequest(
    task="Draft a Twitter launch campaign for my product",
    mode="marketing",      # optional
    platform="api",        # optional
    extra={                 # optional
        "marketing_product": {"product_id": "prod_123"}
    },
)

字段：

• task: str（必填）
• intent_id: Optional[str]
• mode: Optional[str]
• platform: Optional[str]
• extra: Optional[Dict[str, Any]]

TaskRunner API

通过包级工厂函数创建 TaskRunner：

from wyseos.mate import create_task_runner
from wyseos.mate.websocket import WebSocketClient

ws_client = WebSocketClient(
    base_url=client.base_url,
    api_key=client.api_key or "",
    jwt_token=client.jwt_token or "",
    session_id=session_info.session_id,
)

task_runner = create_task_runner(ws_client, client, session_info)

TaskMode

• TaskMode.Default
• TaskMode.DeepSearch
• TaskMode.Marketing

TaskExecutionOptions

TaskRunner 方法

run_task(
    task: str,
    attachments: List[Dict] = None,
    task_mode: TaskMode = TaskMode.Default,
    extra: Optional[Dict[str, Any]] = None,
    options: TaskExecutionOptions = None,
) -> TaskResult

run_interactive_session(
    initial_task: str,
    attachments: List[Dict] = None,
    task_mode: TaskMode = TaskMode.Default,
    extra: Optional[Dict[str, Any]] = None,
    options: TaskExecutionOptions = None,
) -> None

TaskResult

• success: bool
• final_answer: str
• error: Optional[str]
• screenshots: List[Dict[str, Any]]
• execution_logs: List[Dict[str, Any]]
• plan_history: List[Dict[str, Any]]
• session_duration: float
• message_count: int

Product API

通过 client.product 访问。

方法

• create(req: CreateProductRequest) -> CreateProductResponse
• get_info(product_id: str) -> ProductInfo
• get_report(report_id: str) -> ProductReport
• get_categories() -> List[Industry]
• create_and_wait(product: str, attachments: Optional[List[Dict[str, str]]] = None, poll_interval: int = 20, max_attempts: int = 30, on_poll: Optional[Callable[[int, str], None]] = None) -> ProductReport

CreateProductRequest

• product: str
• attachments: List[ProductAttachment]

ProductAttachment：

• file_name: str
• file_url: str

Marketing API

通过 client.marketing 访问。

• update_report(report_id: str, data: Dict[str, Any]) -> Dict[str, Any]
• get_research_tweets(query_id: str) -> Any

会话级营销内容查询：

• client.session.get_marketing_data(session_id, type="reply"|"like"|"retweet"|"tweet")

文件上传 API

通过 client.file_upload 访问。

• upload_file(file_path: str, session_id: Optional[str] = None) -> Dict[str, Any]

支持扩展名：

• .txt, .pdf, .png, .jpg, .jpeg, .md, .csv, .html, .htm, .rss, .xml, .gif, .py, .json

最大文件大小：10MB。

分页模型

• ListOptions(page_num: int = 1, page_size: int = 20)
• PaginatedResponse[T] 字段：page_num、page_size、total、total_page、data