跳转到主要内容

调用智能体

POST /v1/agent/invoke 从你自己的应用或服务向智能体发送一条消息。

身份验证

Authorization 请求头中发送工作区 API 密钥。
Authorization: Bearer YOUR_API_KEY
智能体 webhook 触发器使用智能体 webhook 密钥,而非 API 密钥。请参阅 Webhook 触发器

请求体

在编辑器中打开智能体并点击 Code Snippet,可以在 Fetch Hive 中查看当前的公开请求结构。

Webhook 触发器

POST /v1/agent/webhooks/{agent_id} 当外部服务需要在不使用工作区 API 密钥的情况下启动智能体运行时,使用智能体 webhook 触发器。请求必须在 X-Fetch-Hive-Webhook-Secret 中包含智能体 webhook 密钥。如果第三方工具无法设置自定义请求头,可以退而使用 ?secret=YOUR_WEBHOOK_SECRET 智能体 webhook 触发器始终使用回调投递。包含 async.callback_url;Fetch Hive 返回 202,带有 request_idrun_status: "running",随后向回调 URL 投递 agent.completedagent.failed
curl 'https://api.fetchhive.com/v1/agent/webhooks/YOUR_AGENT_ID' \
  -H 'X-Fetch-Hive-Webhook-Secret: YOUR_WEBHOOK_SECRET' \
  -H 'Content-Type: application/json' \
  --data-raw '{
    "message": "Summarize the latest ticket",
    "thread_id": "external-thread-123",
    "metadata": {
      "customer_id": "cus_123",
      "plan": "enterprise"
    },
    "user": "external-user-1",
    "async": {
      "callback_url": "https://example.com/callback"
    }
  }'
message 是必填项。thread_id 可选,可以重复使用以恢复由调用方管理的线程。metadata 遵循与普通智能体调用相同的扁平且仅含标量的规则,并会出现在日志筛选中。 Fetch Hive 使用与入站触发器认证相同的 webhook 密钥对出站回调进行签名。 回调 dataagent.completed 时包含 response,在 agent.failed 时包含 error;使用顶层的 event_typerequest_id 来表示终态和关联。
字段类型是否必填描述
agentstring智能体 ID
messagestring你想发送的消息
streamingboolean流式传输响应事件,而不是等待一条最终的 JSON 响应
thread_idstring标识会话线程的任意字符串。Fetch Hive 在首次使用时创建新线程,并在后续以相同值调用时恢复该线程。适合持久的多轮对话。
messagesarray作为上下文使用的之前的会话轮次,不会将历史持久化到数据库。每项: { "content": string, "role": "user" | "assistant" | "system", "attachments"?: array, "image_urls"?: array }。当你自行管理会话状态时使用此项。
attachmentsarray附加在当前 message 上的 HTTPS 文件 URL。每项可以是 URL 字符串,也可以是带有 file_url 的对象。受支持的文档类型按扩展名识别为 CSV、XLSX、PDF、DOCX 以及文本/Markdown。此处也可传入图片。
image_urlsarray当前 message 上 HTTPS 图片附件的向后兼容简写。适用于仅图片集成,尤其是当图片 URL 没有扩展名时。
metadataobject调用方定义的扁平元数据,用于审计和日志筛选。不会加入到智能体提示词中。
metadata 必须是扁平对象。键必须是非空字符串,值必须是字符串、数字、布尔值或 null。数组和嵌套对象会在运行启动前被拒绝。 attachments 项可以是简单的 URL 字符串:
[
  "https://example.com/customer-brief.pdf",
  "https://uploads.fetchhive.com/uploads/grme67114iieusl54m3u4swipr3v"
]
当你想提供元数据时,也可以使用如下对象结构:
{
  "file_url": "https://example.com/customer-brief.pdf",
  "file_name": "customer-brief.pdf",
  "file_type": "application/pdf"
}
仅接受 https:// URL。每条消息最多允许五个附件,attachmentsimage_urls 共同计数。文档附件通过系统 read_file 工具以 <available_files> 清单的形式暴露给智能体;智能体在依赖文档内容前会先调用 read_file。如果附件 URL 或类型无效,Fetch Hive 会返回 422 Unprocessable Entity,而不会打开流。 受支持的文档附件:
  • CSV: text/csv, .csv
  • XLSX: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet, .xlsx
  • PDF: application/pdf, .pdf
  • DOCX: application/vnd.openxmlformats-officedocument.wordprocessingml.document, .docx
  • 按扩展名识别的文本和 Markdown: .txt, .md, .markdown
对于没有扩展名的 URL(例如 Fetch Hive 媒体库 URL),请直接传入该 URL。Fetch Hive 会将该 URL 加入 read_file 的允许名单,文件工具在抓取文件时会检测真实类型。 有效示例:
{
  "metadata": {
    "customer_id": "cus_123",
    "plan": "enterprise",
    "trial": false,
    "invoice_count": 12,
    "region": null
  }
}
无效示例:
{
  "metadata": {
    "customer": {
      "id": "cus_123"
    },
    "tags": ["enterprise"]
  }
}
代码片段对话框使用以下请求体结构:
{
  "agent": "AGENT_UUID",
  "message": "Your message here",
  "streaming": true,
  "attachments": [
    "https://example.com/customer-brief.pdf"
  ],
  "metadata": {
    "customer_id": "cus_123",
    "plan": "enterprise"
  }
}
使用 thread_id 进行持久会话时:
{
  "agent": "AGENT_UUID",
  "message": "What did we discuss last time?",
  "streaming": true,
  "thread_id": "user-456-support-session"
}
使用 messages 进行由调用方管理的(无状态)历史时:
{
  "agent": "AGENT_UUID",
  "message": "What else should I know?",
  "streaming": true,
  "messages": [
    {
      "content": "What are the latest AI trends?",
      "role": "user",
      "attachments": [
        "https://example.com/market-data.csv"
      ]
    },
    { "content": "Teams are focusing on evals, routing, and observability.", "role": "assistant" }
  ]
}
如果之前的 assistant 轮次生成了文件,请通过 messages[].attachments 在相应的历史消息中将该文件传回。Fetch Hive 使用当前消息和历史消息中结构化的附件 URL 来为后续轮次授权 read_file 访问;仅在纯文本中提到的 URL 不会被视为文件工具附件。

响应

如果 streamingtrue,Fetch Hive 返回事件流。如果在流打开后提供商失败,Fetch Hive 会在关闭流前发送一个最终的 error 事件。 推理事件示例:
{
  "request_id": "req_019d52846ea37682b03522fd0695cc43",
  "type": "reasoning",
  "response": "Looking at the latest model releases..."
}
响应事件示例:
{
  "request_id": "req_019d52846ea37682b03522fd0695cc43",
  "type": "response",
  "response": "Teams are standardizing around evals, routing, and observability.",
  "done": false
}
工具事件示例:
{
  "request_id": "req_019d52846ea37682b03522fd0695cc43",
  "type": "tool",
  "tool_id": "tool_123",
  "tool": "google_search",
  "tool_input": {
    "query": "latest AI infrastructure trends 2026"
  },
  "observation": {
    "results": []
  }
}
最终 usage 事件示例:
{
  "request_id": "req_019d52846ea37682b03522fd0695cc43",
  "type": "usage",
  "usage": {
    "duration": 4.79230260848999,
    "input_tokens": {
      "total_tokens": 24,
      "cached_tokens": 0
    },
    "output_tokens": {
      "total_tokens": 170,
      "reasoning_tokens": 64
    },
    "total_tokens": 194
  },
  "stop_reason": "completed"
}
错误事件示例:
{
  "request_id": "req_019d52846ea37682b03522fd0695cc43",
  "type": "error",
  "id": "msg_019d52846ea37682b03522fd0695cc43",
  "error": "openai provider stream error: HTTP 400: invalid request",
  "message": "openai provider stream error: HTTP 400: invalid request",
  "provider": "openai",
  "error_type": "provider_stream_error",
  "status_code": 502
}
如果 streamingfalse,Fetch Hive 返回一条 JSON 响应。提供商执行失败时返回 502 Bad Gateway,带有 error 消息。
{
  "request_id": "req_019d528660dd7e22b15e5b13a1931c50",
  "model": "gpt-5-nano-2025-08-07",
  "duration": 4.641960144042969,
  "response": "Teams are moving from simple wrappers to systems with evals, tool routing, and tighter cost controls.",
  "reasoning": "The request asks for a short summary of current infrastructure trends.",
  "usage": {
    "input_tokens": {
      "total_tokens": 24,
      "cached_tokens": 0
    },
    "output_tokens": {
      "total_tokens": 187,
      "reasoning_tokens": 64
    },
    "total_tokens": 211
  },
  "stop_reason": "completed"
}

示例

curl 'https://api.fetchhive.com/v1/agent/invoke' \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json' \
  --data-raw '{
    "agent": "AGENT_UUID",
    "message": "Summarize the latest AI infrastructure trends",
    "metadata": {
      "customer_id": "cus_123",
      "plan": "enterprise"
    },
    "streaming": true
  }' \
  --compressed

相关内容