360 Secure SkillHub
360 安全技能中心

Acedatacloud Api

面向开发者的 AceDataCloud AI 数据服务 API 使用说明,涵盖账号与凭证准备、Bearer Token 认证、同步与异步请求模式、错误码处理、计费规则,以及 OpenAI 兼容 SDK 的 Python 与 JavaScript 调用示例。

Germey
Germey
热度 52.54
最近更新 2026-05-11 17:34:59
分类:开发依赖:第三方 API应用场景:API 集成

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add acedatacloud-acedatacloud-api
Metadata
name
acedatacloud-api
description
Guide for using AceDataCloud APIs. Use when authenticating, making API calls, managing credentials, understanding billing, or integrating AceDataCloud services into applications. Covers setup, authentication, request patterns, error handling, and SDK integration.
license
Apache-2.0
metadata
{"author":"acedatacloud","version":"1.0"}
compatibility
Requires an AceDataCloud account at platform.acedata.cloud.

AceDataCloud API Usage Guide

Complete guide for using AceDataCloud's AI-powered data services API.

Getting Started

1. Create an Account

Register at platform.acedata.cloud.

2. Subscribe to a Service

Browse available services and click Get to subscribe. Most services include free quota.

3. Create API Credentials

Go to your service's Credentials page and create an API Token.

Authentication

All APIs use Bearer token authentication:

curl -X POST https://api.acedata.cloud/<endpoint> \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ ... }'

Token Types

Type Format Scope
Service Token UUID string Access to subscribed service only
Global Token UUID string Access to all subscribed services

Base URL

https://api.acedata.cloud

SDK Integration (OpenAI-Compatible)

For chat completion services, use the standard OpenAI SDK:

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_TOKEN",
    base_url="https://api.acedata.cloud/v1"
)

response = client.chat.completions.create(
    model="claude-sonnet-4-20250514",
    messages=[{"role": "user", "content": "Hello!"}]
)
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "YOUR_API_TOKEN",
  baseURL: "https://api.acedata.cloud/v1"
});

const response = await client.chat.completions.create({
  model: "gpt-4.1",
  messages: [{ role: "user", content: "Hello!" }]
});

Common Request Patterns

Synchronous APIs

Some APIs return results immediately (e.g., face transform, search):

curl -X POST https://api.acedata.cloud/face/analyze \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"image_url": "https://example.com/photo.jpg"}'

Async Task APIs

Most generation APIs (images, video, music) are asynchronous:

Step 1: Submit request

POST /suno/audios
{"prompt": "a jazz song", "wait": false}
→ {"task_id": "abc123"}

Step 2: Poll for results

POST /suno/tasks
{"task_id": "abc123"}
→ {"state": "succeeded", "data": [...]}

Shortcut: Pass "wait": true to block until completion (simpler but longer timeout).

Error Handling

HTTP Code Meaning Action
400 Bad request Check request parameters
401 Unauthorized Check API token
403 Forbidden Content filtered or insufficient permissions
429 Rate limited Wait and retry with backoff
500 Server error Retry or contact support

Error response format:

{
  "error": {
    "code": "token_mismatched",
    "message": "Invalid or expired token"
  }
}

Billing

  • Each API call deducts from your subscription balance (remaining_amount)
  • Cost varies by service, model, and usage (tokens, requests, data size)
  • Check balance at platform.acedata.cloud
  • Most services offer free trial quota

Service Categories

Category Services Base Path
AI Chat GPT, Claude, Gemini, DeepSeek, Grok /v1/chat/completions
Image Gen Midjourney, Flux, Seedream, NanoBanana /midjourney/*, /flux/*, etc.
Video Gen Luma, Sora, Veo, Kling, Hailuo, Seedance /luma/*, /sora/*, etc.
Music Gen Suno, Producer, Fish Audio /suno/*, /producer/*, /fish/*
Search Google Search (web/images/news/maps) /serp/*
Face Analyze, beautify, swap, cartoon, age /face/*
Utility Short URL, QR Art, Headshots /short-url, /qrart/*, /headshots/*

Gotchas

  • Tokens are service-scoped by default — create a global token if you need cross-service access
  • Async APIs return a task_id — you must poll to get the result
  • wait: true is convenient but has a timeout limit (typically 60–120s)
  • Rate limits vary by service tier — upgrade your plan if hitting limits
  • All timestamps are in UTC