Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Decentralized compute and data marketplace for AI agents with spot pricing | 去中心化 AI Agent 计算和数据市场,支持 Spot 动态定价
.claude/skills/leoyeai-decentralized-agent-cloud/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 503% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 245% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 226% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 121% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 200% | 0% |
> Decentralized compute and data marketplace for AI agents with spot pricing > > 为 AI Agent 提供去中心化计算和数据市场,支持动态 Spot 定价
English: A peer-to-peer marketplace where AI agents can discover and execute computational skills (video generation, data processing, ML inference, etc.) with real-time spot pricing. Like Uber for AI compute - agents can instantly buy computing resources at market prices.
中文: 一个点对点的市场平台,AI Agent 可以发现和执行计算技能(视频生成、数据处理、机器学习推理等),使用实时 Spot 定价。就像 Uber 一样,但用于 AI 计算 - Agent 可以按市场价格即时购买计算资源。
English:
bash# Via npm npm install openclaw-decentralized-agent-cloud # Via ClawHub clawhub install decentralized-agent-cloud
中文:
bash# 通过 npm npm install openclaw-decentralized-agent-cloud # 通过 ClawHub clawhub install decentralized-agent-cloud
English:
typescriptimport { createAgentClient } from 'openclaw-decentralized-agent-cloud'; // Initialize client const client = createAgentClient({ apiKey: 'your-api-key', agentId: 'my-agent', }); // Execute video generation skill const task = await client.executeSkill('video-generator', { script: 'AI is changing the world. Three breakthroughs happened today.', voice: 'nova', speed: 1.15, }); // Wait for result const result = await client.waitForTask(task.id); console.log(result.output.videoUrl); // Output: https://storage.agent-cloud.io/videos/abc123.mp4
中文:
typescriptimport { createAgentClient } from 'openclaw-decentralized-agent-cloud'; // 初始化客户端 const client = createAgentClient({ apiKey: 'your-api-key', agentId: 'my-agent', }); // 执行视频生成技能 const task = await client.executeSkill('video-generator', { script: 'AI 正在改变世界。今天发生了三个突破。', voice: 'nova', speed: 1.15, }); // 等待结果 const result = await client.waitForTask(task.id); console.log(result.output.videoUrl); // 输出: https://storage.agent-cloud.io/videos/abc123.mp4
1. AI Video Generation
typescript// Generate professional short videos from text const video = await client.executeSkill('video-generator', { script: 'Your marketing message here...', voice: 'nova', });
2. Data Processing
typescript// Process large datasets with auto-scaling compute const result = await client.executeSkill('data-processor', { dataset: 's3://my-bucket/data.csv', operation: 'analyze', resourceRequirements: { compute: { type: 'GPU', memory: 16 } } });
3. ML Model Inference
typescript// Run inference on available GPU clusters const prediction = await client.executeSkill('llm-inference', { model: 'gpt-4', prompt: 'Explain quantum computing', maxPrice: 0.50 // bid up to $0.50/hour });
1. AI 视频生成
typescript// 从文本生成专业短视频 const video = await client.executeSkill('video-generator', { script: '您的营销信息...', voice: 'nova', });
2. 数据处理
typescript// 使用自动扩展计算处理大型数据集 const result = await client.executeSkill('data-processor', { dataset: 's3://my-bucket/data.csv', operation: 'analyze', resourceRequirements: { compute: { type: 'GPU', memory: 16 } } });
3. ML 模型推理
typescript// 在可用的 GPU 集群上运行推理 const prediction = await client.executeSkill('llm-inference', { model: 'gpt-4', prompt: '解释量子计算', maxPrice: 0.50 // 出价最高 $0.50/小时 });
Traditional Cloud (AWS):
Decentralized Agent Cloud:
| Resource | Traditional | Spot Price | Savings | |----------|------------|------------|---------| | CPU (4 cores, 8GB) | $0.20/hr | $0.05/hr | 75% | | GPU (V100 16GB) | $3.06/hr | $0.50/hr | 84% | | GPU (A100 40GB) | $5.50/hr | $1.20/hr | 78% |
传统云(AWS):
去中心化 Agent 云:
| 资源 | 传统云 | Spot 价格 | 节省 | |------|--------|-----------|------| | CPU (4核, 8GB) | $0.20/时 | $0.05/时 | 75% | | GPU (V100 16GB) | $3.06/时 | $0.50/时 | 84% | | GPU (A100 40GB) | $5.50/时 | $1.20/时 | 78% |
┌─────────────────────────────────────────────────┐
│ DECENTRALIZED AGENT CLOUD │
│ │
│ ┌──────────────┐ ┌──────────────┐ │
│ │Skill Registry│ │Spot Pricing │ │
│ │ │ │ Engine │ │
│ │• Discovery │ │• Supply/ │ │
│ │• Validation │ │ Demand │ │
│ └──────────────┘ └──────────────┘ │
│ │
│ ┌──────────────┐ ┌──────────────┐ │
│ │Task Scheduler│ │Resource │ │
│ │ │ │Marketplace │ │
│ │• Matching │ │• Providers │ │
│ │• Execution │ │• Consumers │ │
│ └──────────────┘ └──────────────┘ │
└─────────────────────────────────────────────────┘
▲ ▲
│ │
┌────┴────┐ ┌─────┴─────┐
│ Agents │ │ Providers │
│(Buyers) │ │ (Sellers) │
└─────────┘ └───────────┘Core Components:
┌─────────────────────────────────────────────────┐
│ 去中心化 AGENT 云 │
│ │
│ ┌──────────────┐ ┌──────────────┐ │
│ │技能注册表 │ │Spot 定价 │ │
│ │ │ │ 引擎 │ │
│ │• 发现 │ │• 供需 │ │
│ │• 验证 │ │ 关系 │ │
│ └──────────────┘ └──────────────┘ │
│ │
│ ┌──────────────┐ ┌──────────────┐ │
│ │任务调度器 │ │资源 │ │
│ │ │ │市场 │ │
│ │• 匹配 │ │• 提供者 │ │
│ │• 执行 │ │• 消费者 │ │
│ └──────────────┘ └──────────────┘ │
└─────────────────────────────────────────────────┘
▲ ▲
│ │
┌────┴────┐ ┌─────┴─────┐
│ Agent │ │ 提供者 │
│(买家) │ │ (卖家) │
└─────────┘ └───────────┘核心组件:
| Category | Skill | Description | Base Price | |----------|-------|-------------|------------| | Video | video-generator | Text-to-video with TTS + effects | $0.10/video | | Audio | transcription | Speech-to-text (Whisper) | $0.006/min | | Audio | tts-synthesis | Text-to-speech (11 voices) | $0.015/1K chars | | Image | image-generator | AI image generation (DALL-E) | $0.02/image | | Text | llm-inference | LLM inference (GPT-4, Claude) | $0.50/hour | | ML | model-training | Custom model training | Spot price | | Data | data-pipeline | ETL and analytics | Spot price |
| 类别 | 技能 | 描述 | 基础价格 | |------|------|------|----------| | 视频 | video-generator | 文本转视频(TTS + 特效) | $0.10/视频 | | 音频 | transcription | 语音转文本(Whisper) | $0.006/分钟 | | 音频 | tts-synthesis | 文本转语音(11种声音) | $0.015/1K字符 | | 图像 | image-generator | AI 图像生成(DALL-E) | $0.02/图像 | | 文本 | llm-inference | LLM 推理(GPT-4, Claude) | $0.50/小时 | | ML | model-training | 自定义模型训练 | Spot 价格 | | 数据 | data-pipeline | ETL 和分析 | Spot 价格 |
Skills Executable capabilities (video generation, ML inference, data processing) that agents can discover and execute.
Spot Pricing Dynamic pricing based on real-time supply and demand. Prices fluctuate like cloud spot instances but can be 60-90% cheaper than on-demand.
Task Scheduler Matches tasks with providers based on:
Resource Providers Anyone can monetize idle compute by registering as a provider. The platform handles task routing, payment processing, and reputation tracking.
技能 可执行的能力(视频生成、ML 推理、数据处理),Agent 可以发现和执行。
Spot 定价 基于实时供需的动态定价。价格像云 Spot 实例一样波动,但比按需定价便宜 60-90%。
任务调度器 根据以下条件将任务匹配到提供者:
资源提供者 任何人都可以通过注册为提供者来变现闲置计算资源。平台处理任务路由、支付处理和声誉跟踪。
English:
typescriptimport { skillRegistry } from 'openclaw-decentralized-agent-cloud/core'; // Define your skill const mySkill = { id: 'image-classifier', name: 'Image Classification', version: '1.0.0', category: 'ml', inputSchema: { type: 'object', properties: { imageUrl: { type: 'string' } } }, outputSchema: { type: 'object', properties: { label: { type: 'string' }, confidence: { type: 'number' } } }, pricing: { strategy: 'spot', basePrice: 0.05 } }; // Register await skillRegistry.registerSkill(mySkill);
中文:
typescriptimport { skillRegistry } from 'openclaw-decentralized-agent-cloud/core'; // 定义你的技能 const mySkill = { id: 'image-classifier', name: '图像分类', version: '1.0.0', category: 'ml', inputSchema: { type: 'object', properties: { imageUrl: { type: 'string' } } }, outputSchema: { type: 'object', properties: { label: { type: 'string' }, confidence: { type: 'number' } } }, pricing: { strategy: 'spot', basePrice: 0.05 } }; // 注册 await skillRegistry.registerSkill(mySkill);
English:
typescriptimport { taskScheduler } from 'openclaw-decentralized-agent-cloud/core'; // Register your compute resources await taskScheduler.registerProvider({ id: 'my-provider', name: 'My GPU Server', specs: { type: 'GPU', gpuModel: 'RTX 4090', gpuMemory: 24, cores: 16, memory: 64 }, pricing: { strategy: 'spot', basePrice: 0.80 // USD/hour } });
中文:
typescriptimport { taskScheduler } from 'openclaw-decentralized-agent-cloud/core'; // 注册你的计算资源 await taskScheduler.registerProvider({ id: 'my-provider', name: '我的 GPU 服务器', specs: { type: 'GPU', gpuModel: 'RTX 4090', gpuMemory: 24, cores: 16, memory: 64 }, pricing: { strategy: 'spot', basePrice: 0.80 // 美元/小时 } });
For Agents (Buyers)
For Providers (Sellers)
For Platform
对 Agent(买家)
对提供者(卖家)
对平台
examples/complete-demo.tsexamples/complete-demo.tsPhase 1: Core Platform (Current) ✅
Phase 2: Enhanced Features (Q2 2026) 🚧
Phase 3: Full Decentralization (Q4 2026) 🔮
阶段 1:核心平台(当前) ✅
阶段 2:增强功能(2026年第2季度) 🚧
阶段 3:完全去中心化(2026年第4季度) 🔮
MIT License
English: We welcome contributions! Create new skills, improve documentation, or add features.
中文: 我们欢迎贡献!创建新技能、改进文档或添加功能。
Built with ❤️ for the autonomous agent economy
为自主 Agent 经济构建 ❤️
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,479 | 4,910 | -24% | 1 | 1 | 0% | 1,550 | 5,884 | +280% | 0 | 0 | — |
case-02 | fail→pass | 25,945 | 13,848 | -47% | 1 | 1 | 0% | 1,345 | 8,111 | +503% | 0 | 0 | — |
case-03 | fail→fail | 12,645 | 7,917 | -37% | 1 | 1 | 0% | 2,661 | 6,434 | +142% | 0 | 0 | — |
case-04 | fail→fail | 8,374 | 10,863 | +30% | 1 | 1 | 0% | 1,619 | 6,532 | +303% | 0 | 0 | — |
case-05 | pass→pass | 11,231 | 11,959 | +6% | 1 | 1 | 0% | 2,308 | 6,923 | +200% | 0 | 0 | — |
case-06 | fail→fail | 16,637 | 12,728 | -23% | 1 | 1 | 0% | 3,129 | 7,441 | +138% | 0 | 0 | — |
case-07 | pass→pass | 2,651 | 2,462 | -7% | 1 | 1 | 0% | 560 | 5,180 | +825% | 0 | 0 | — |
case-08 | fail→fail | 12,513 | 13,242 | +6% | 1 | 1 | 0% | 2,468 | 7,514 | +204% | 0 | 0 | — |
case-09 | pass→pass | 13,883 | 16,318 | +18% | 1 | 1 | 0% | 2,682 | 7,904 | +195% | 0 | 0 | — |
case-10 | fail→pass | 10,565 | 8,291 | -22% | 1 | 1 | 0% | 1,841 | 6,343 | +245% | 0 | 0 | — |
case-11 | fail→fail | 16,282 | 14,785 | -9% | 1 | 1 | 0% | 3,237 | 7,642 | +136% | 0 | 0 | — |
case-12 | fail→fail | 4,516 | 5,759 | +28% | 1 | 1 | 0% | 799 | 5,842 | +631% | 0 | 0 | — |
case-13 | fail→fail | 8,328 | 6,253 | -25% | 1 | 1 | 0% | 1,871 | 6,092 | +226% | 0 | 0 | — |
case-14 | fail→fail | 15,450 | 20,967 | +36% | 1 | 1 | 0% | 2,771 | 8,408 | +203% | 0 | 0 | — |
case-15 | fail→pass | 12,062 | 11,511 | -5% | 1 | 1 | 0% | 2,153 | 7,019 | +226% | 0 | 0 | — |
case-16 | pass→pass | 10,389 | 11,572 | +11% | 1 | 1 | 0% | 2,108 | 6,843 | +225% | 0 | 0 | — |
case-17 | fail→fail | 13,223 | 13,662 | +3% | 1 | 1 | 0% | 2,508 | 7,636 | +204% | 0 | 0 | — |
case-18 | fail→fail | 12,120 | 12,389 | +2% | 1 | 1 | 0% | 2,297 | 7,126 | +210% | 0 | 0 | — |
case-19 | pass→pass | 12,114 | 12,627 | +4% | 1 | 1 | 0% | 2,225 | 6,983 | +214% | 0 | 0 | — |
case-20 | fail→pass | 16,564 | 11,992 | -28% | 1 | 1 | 0% | 3,494 | 7,729 | +121% | 0 | 0 | — |
case-21 | pass→pass | 10,153 | 9,534 | -6% | 1 | 1 | 0% | 2,103 | 6,704 | +219% | 0 | 0 | — |
case-22 | pass→pass | 7,086 | 8,944 | +26% | 1 | 1 | 0% | 1,534 | 6,633 | +332% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.