javascript-sdk

安装 belt CLI 技能： npx skills add belt-sh/cli

JavaScript SDK

使用 inference.sh JavaScript/TypeScript SDK 构建 AI 应用。

快速入门

npm install @inferencesh/sdk

import { inference } from '@inferencesh/sdk';

const client = inference({ apiKey: 'inf_your_key' });

// Run an AI app
const result = await client.run({
  app: 'infsh/flux-schnell',
  input: { prompt: 'A sunset over mountains' }
});
console.log(result.output);

安装

npm install @inferencesh/sdk
# or
yarn add @inferencesh/sdk
# or
pnpm add @inferencesh/sdk

要求： Node.js 18.0.0+（或支持 fetch 的现代浏览器）

身份验证

import { inference } from '@inferencesh/sdk';

// Direct API key
const client = inference({ apiKey: 'inf_your_key' });

// From environment variable (recommended)
const client = inference({ apiKey: process.env.INFERENCE_API_KEY });

// For frontend apps (use proxy)
const client = inference({ proxyUrl: '/api/inference/proxy' });

获取 API 密钥：Settings → API Keys → Create API Key

运行应用

基本执行

const result = await client.run({
  app: 'infsh/flux-schnell',
  input: { prompt: 'A cat astronaut' }
});

console.log(result.status);  // "completed"
console.log(result.output);  // Output data

启动后不等待结果

const task = await client.run({
  app: 'google/veo-3-1-fast',
  input: { prompt: 'Drone flying over mountains' }
}, { wait: false });

console.log(`Task ID: ${task.id}`);
// Check later with client.getTask(task.id)

流式传输进度

const stream = await client.run({
  app: 'google/veo-3-1-fast',
  input: { prompt: 'Ocean waves at sunset' }
}, { stream: true });

for await (const update of stream) {
  console.log(`Status: ${update.status}`);
  if (update.logs?.length) {
    console.log(update.logs.at(-1));
  }
}

运行参数

参数	类型	描述
app	string	App ID（namespace/name@version）
input	object	与应用架构匹配的输入
setup	object	隐藏的设置配置
infra	string	'cloud' 或 'private'
session	string	有状态执行的会话 ID
session_timeout	number	空闲超时（1-3600 秒）

文件处理

自动上传

const result = await client.run({
  app: 'image-processor',
  input: {
    image: '/path/to/image.png'  // Auto-uploaded
  }
});

手动上传

// Basic upload
const file = await client.uploadFile('/path/to/image.png');

// With options
const file = await client.uploadFile('/path/to/image.png', {
  filename: 'custom_name.png',
  contentType: 'image/png',
  public: true
});

const result = await client.run({
  app: 'image-processor',
  input: { image: file.uri }
});

浏览器文件上传

const input = document.querySelector('input[type="file"]');
const file = await client.uploadFile(input.files[0]);

会话（有状态执行）

在多次调用期间保持工作进程处于预热状态：

// Start new session
const result = await client.run({
  app: 'my-app',
  input: { action: 'init' },
  session: 'new',
  session_timeout: 300  // 5 minutes
});
const sessionId = result.session_id;

// Continue in same session
const result2 = await client.run({
  app: 'my-app',
  input: { action: 'process' },
  session: sessionId
});

Agent SDK

模板 Agent

使用工作区中的预构建 Agent：

const agent = client.agent('my-team/support-agent@latest');

// Send message
const response = await agent.sendMessage('Hello!');
console.log(response.text);

// Multi-turn conversation
const response2 = await agent.sendMessage('Tell me more');

// Reset conversation
agent.reset();

// Get chat history
const chat = await agent.getChat();

临时 Agent

以编程方式创建自定义 Agent：

import { tool, string, number, appTool } from '@inferencesh/sdk';

// Define tools
const calculator = tool('calculate')
  .describe('Perform a calculation')
  .param('expression', string('Math expression'))
  .build();

const imageGen = appTool('generate_image', 'infsh/flux-schnell@latest')
  .describe('Generate an image')
  .param('prompt', string('Image description'))
  .build();

// Create agent
const agent = client.agent({
  core_app: { ref: 'infsh/claude-sonnet-4@latest' },
  system_prompt: 'You are a helpful assistant.',
  tools: [calculator, imageGen],
  temperature: 0.7,
  max_tokens: 4096
});

const response = await agent.sendMessage('What is 25 * 4?');

可用的核心应用

模型	应用引用
Claude Sonnet 4	infsh/claude-sonnet-4@latest
Claude 3.5 Haiku	infsh/claude-haiku-35@latest
GPT-4o	infsh/gpt-4o@latest
GPT-4o Mini	infsh/gpt-4o-mini@latest

工具构建器 API

参数类型

import {
  string, number, integer, boolean,
  enumOf, array, obj, optional
} from '@inferencesh/sdk';

const name = string('User\'s name');
const age = integer('Age in years');
const score = number('Score 0-1');
const active = boolean('Is active');
const priority = enumOf(['low', 'medium', 'high'], 'Priority');
const tags = array(string('Tag'), 'List of tags');
const address = obj({
  street: string('Street'),
  city: string('City'),
  zip: optional(string('ZIP'))
}, 'Address');

客户端工具（在你的代码中运行）

const greet = tool('greet')
  .display('Greet User')
  .describe('Greets a user by name')
  .param('name', string('Name to greet'))
  .requireApproval()
  .build();

应用工具（调用 AI 应用）

const generate = appTool('generate_image', 'infsh/flux-schnell@latest')
  .describe('Generate an image from text')
  .param('prompt', string('Image description'))
  .setup({ model: 'schnell' })
  .input({ steps: 20 })
  .requireApproval()
  .build();

Agent 工具（委派给子 Agent）

import { agentTool } from '@inferencesh/sdk';

const researcher = agentTool('research', 'my-org/researcher@v1')
  .describe('Research a topic')
  .param('topic', string('Topic to research'))
  .build();

Webhook 工具（调用外部 API）

import { webhookTool } from '@inferencesh/sdk';

const notify = webhookTool('slack', 'https://hooks.slack.com/...')
  .describe('Send Slack notification')
  .secret('SLACK_SECRET')
  .param('channel', string('Channel'))
  .param('message', string('Message'))
  .build();

内部工具（内置功能）

import { internalTools } from '@inferencesh/sdk';

const config = internalTools()
  .plan()
  .memory()
  .webSearch(true)
  .codeExecution(true)
  .imageGeneration({
    enabled: true,
    appRef: 'infsh/flux@latest'
  })
  .build();

const agent = client.agent({
  core_app: { ref: 'infsh/claude-sonnet-4@latest' },
  internal_tools: config
});

流式传输 Agent 响应

const response = await agent.sendMessage('Explain quantum computing', {
  onMessage: (msg) => {
    if (msg.content) {
      process.stdout.write(msg.content);
    }
  },
  onToolCall: async (call) => {
    console.log(`\n[Tool: ${call.name}]`);
    const result = await executeTool(call.name, call.args);
    agent.submitToolResult(call.id, result);
  }
});

文件附件

// From file path (Node.js)
import { readFileSync } from 'fs';
const response = await agent.sendMessage('What\'s in this image?', {
  files: [readFileSync('image.png')]
});

// From base64
const response = await agent.sendMessage('Analyze this', {
  files: ['data:image/png;base64,iVBORw0KGgo...']
});

// From browser File object
const input = document.querySelector('input[type="file"]');
const response = await agent.sendMessage('Describe this', {
  files: [input.files[0]]
});

技能（可复用上下文）

const agent = client.agent({
  core_app: { ref: 'infsh/claude-sonnet-4@latest' },
  skills: [
    {
      name: 'code-review',
      description: 'Code review guidelines',
      content: '# Code Review\n\n1. Check security\n2. Check performance...'
    },
    {
      name: 'api-docs',
      description: 'API documentation',
      url: 'https://example.com/skills/api-docs.md'
    }
  ]
});

服务代理（前端应用）

对于浏览器应用，可通过后端进行代理，以确保 API 密钥安全：

客户端设置

const client = inference({
  proxyUrl: '/api/inference/proxy'
  // No apiKey needed on frontend
});

Next.js 代理（App Router）

// app/api/inference/proxy/route.ts
import { createRouteHandler } from '@inferencesh/sdk/proxy/nextjs';

const route = createRouteHandler({
  apiKey: process.env.INFERENCE_API_KEY
});

export const POST = route.POST;

Express 代理

import express from 'express';
import { createProxyMiddleware } from '@inferencesh/sdk/proxy/express';

const app = express();
app.use('/api/inference/proxy', createProxyMiddleware({
  apiKey: process.env.INFERENCE_API_KEY
}));

支持的框架

• Next.js（App Router 和 Pages Router）
• Express
• Hono
• Remix
• SvelteKit

TypeScript 支持

包含完整的类型定义：

import type {
  TaskDTO,
  ChatDTO,
  ChatMessageDTO,
  AgentTool,
  TaskStatusCompleted,
  TaskStatusFailed
} from '@inferencesh/sdk';

if (result.status === TaskStatusCompleted) {
  console.log('Done!');
} else if (result.status === TaskStatusFailed) {
  console.log('Failed:', result.error);
}

错误处理

import { RequirementsNotMetException, InferenceError } from '@inferencesh/sdk';

try {
  const result = await client.run({ app: 'my-app', input: {...} });
} catch (e) {
  if (e instanceof RequirementsNotMetException) {
    console.log('Missing requirements:');
    for (const err of e.errors) {
      console.log(`  - ${err.type}: ${err.key}`);
    }
  } else if (e instanceof InferenceError) {
    console.log('API error:', e.message);
  }
}

人工审批工作流

const response = await agent.sendMessage('Delete all temp files', {
  onToolCall: async (call) => {
    if (call.requiresApproval) {
      const approved = await promptUser(`Allow ${call.name}?`);
      if (approved) {
        const result = await executeTool(call.name, call.args);
        agent.submitToolResult(call.id, result);
      } else {
        agent.submitToolResult(call.id, { error: 'Denied by user' });
      }
    }
  }
});

CommonJS 支持

const { inference, tool, string } = require('@inferencesh/sdk');

const client = inference({ apiKey: 'inf_...' });
const result = await client.run({...});

参考文件

• Agent 模式 - 多 Agent、RAG、批处理模式
• 工具构建器 - 完整的工具构建器 API 参考
• 服务代理 - Next.js、Express、Hono、Remix、SvelteKit 设置
• 流式传输 - 实时进度更新和 SSE 处理
• 文件处理 - 上传、下载和管理文件
• 会话 - 使用预热工作进程进行有状态执行
• TypeScript - 类型定义和类型安全模式
• React 集成 - Hook、组件和模式

相关技能

# Python SDK
npx skills add inference-sh/skills@python-sdk

# Full platform skill (all apps via CLI)
npx skills add inference-sh/skills@infsh-cli

# LLM models
npx skills add inference-sh/skills@llm-models

# Image generation
npx skills add inference-sh/skills@ai-image-generation

文档

• JavaScript SDK 参考 - 完整的 API 文档
• Agent SDK 概览 - 构建 Agent
• 工具构建器参考 - 创建工具
• 服务代理设置 - 前端集成
• 身份验证 - API 密钥设置
• 流式传输 - 实时更新
• 文件上传 - 文件处理