经 AI Skill Hub 精选评估,PulseStack 获评「推荐使用」。这款Agent工作流在功能完整性、社区活跃度和易用性方面表现出色,AI 评分 7.5 分,适合有一定技术背景的用户使用。
PulseStack 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
PulseStack 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:npm 全局安装 npm install -g pulsestack # 方式二:npx 直接运行(无需安装) npx pulsestack --help # 方式三:项目依赖安装 npm install pulsestack # 方式四:从源码运行 git clone https://github.com/NEXARA-oss/PulseStack cd PulseStack npm install npm start
# 命令行使用
pulsestack --help
# 基本用法
pulsestack [options] <input>
# Node.js 代码中使用
const pulsestack = require('pulsestack');
const result = await pulsestack.run(options);
console.log(result);
# pulsestack 配置说明 # 查看配置选项 pulsestack --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export PULSESTACK_CONFIG="/path/to/config.yml"
---
| Feature | Description |
|---|---|
| 🤖 **AI Workflow Runtime** | Execute distributed AI workflows |
| 🔄 **Distributed Event Pipeline** | Real-time event streaming across services |
| 🔁 **Deterministic Replay Engine** | Replay workflow executions for debugging |
| 🌐 **OpenTelemetry Tracing** | End-to-end execution visibility |
| ⚡ **DAG Execution Engine** | Dependency-aware workflow orchestration |
| 🔌 **Plugin System** | Custom runtime extensions and integrations |
| 📊 **Runtime Analytics** | Latency, retries, tokens, throughput, failures |
| 🖥️ **React Operations Console** | Visual workflow and trace inspection |
| ☸️ **Kubernetes + Helm** | Production-ready deployment support |
---
Before getting started, ensure the following tools are installed:
---
pnpm install
Create a .env file from the example file:
Ensure Docker Desktop is running before executing:
docker compose -f infra/docker/docker-compose.yml up -d
Ensure Docker Desktop is installed and running before starting infrastructure services.
The following example demonstrates a simple AI workflow consisting of an Agent → Tool → LLM execution chain.
id="qomqyi"
Plan Agent
│
▼
Fetch Logs
│
▼
Summarize with LLM
id="y2wcfb"
{
"workflow": {
"id": "wf_agent_ops",
"name": "Agent Ops",
"version": "1.0.0",
"tenantId": "local",
"correlationId": "corr_agent_ops",
"metadata": {},
"steps": [
{
"id": "s1",
"name": "Plan",
"kind": "agent",
"dependsOn": [],
"input": {
"objective": "inspect queue health"
}
},
{
"id": "s2",
"name": "FetchLogs",
"kind": "tool",
"dependsOn": ["s1"],
"retry": {
"maxAttempts": 3,
"backoffMs": 250,
"maxBackoffMs": 2000,
"exponential": true
},
"input": {
"tool": "logs.query",
"range": "15m"
}
},
{
"id": "s3",
"name": "Summarize",
"kind": "llm",
"dependsOn": ["s2"],
"input": {
"model": "gpt-4.1",
"prompt": "Summarize anomalies"
}
}
]
},
"input": {
"environment": "prod"
},
"initiatedBy": "operator"
}
```bash pnpm dev
| Variable | Purpose | Default Value |
|---|---|---|
| RUNTIME_URL | Runtime engine service | http://localhost:4101 |
| EVENTS_URL | Event streaming service | http://localhost:4102 |
| TRACE_URL | Tracing service | http://localhost:4103 |
| REPLAY_URL | Replay engine | http://localhost:4104 |
| METRICS_URL | Metrics service | http://localhost:4105 |
| GRAPH_URL | Workflow graph service | http://localhost:4106 |
| VITE_GATEWAY_URL | Frontend gateway URL | http://localhost:4000 |
A sample .env.example file is included with default local development values.
PulseStack是一个开源的AI工作流平台,提供了可观察性和运行时智能功能,但其星数较少,可能需要进一步优化
该工具未明确声明开源协议,商业使用前请联系原作者确认授权范围,避免侵权风险。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
AI Skill Hub 点评:PulseStack 的核心功能完整,质量良好。对于自动化工程师和运维人员来说,这是一个值得纳入个人工具库的选择。建议先在非生产环境试用,再逐步推广。
| 原始名称 | PulseStack |
| 原始描述 | 开源AI工作流:Open-source observability and runtime intelligence platform for distributed AI w。⭐6 · TypeScript |
| Topics | workflowgssoc26typescript |
| GitHub | https://github.com/NEXARA-oss/PulseStack |
| 语言 | TypeScript |
收录时间:2026-06-08 · 更新时间:2026-06-08 · License:未公布 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端