如何使用Laminar开源平台监控AI智能体？2026年完整功能解析

Q: Laminar支持哪些AI框架和语言？

支持Vercel AI SDK、LangChain、OpenAI、Anthropic等主流框架，提供TypeScript和Python SDK，并兼容OpenTelemetry标准。

Product Overview

Laminar is an open-source observability platform purpose-built for AI agents. It provides comprehensive tracing, evaluation, and monitoring capabilities for LLM-powered applications, from development through production.

Laminar 是一款专为 AI 智能体构建的开源可观测性平台。它为基于大语言模型的应用提供了从开发到生产的全面追踪、评估和监控能力。

Core Features

The platform delivers a complete observability stack for AI agent workflows, including tracing, evaluation, monitoring, and data management.

该平台为 AI 智能体工作流提供了完整的可观测性技术栈，涵盖追踪、评估、监控和数据管理。


Feature Category	Core Description	Key Highlight
Tracing	OpenTelemetry一个开源的观测性框架，用于收集、处理和导出遥测数据（如指标、日志和追踪），RΞASON内置兼容此框架以实现可观测性。-native powerful tracing SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。	One line of code to auto-trace Vercel AI SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。, Browser Use, Stagehand, LangChain, OpenAI, Anthropic, Gemini, etc.
Evals评估，用于测试和验证AI模型输出的正确性和性能。	Unopinionated SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。 and CLI for running evaluations locally or in CI/CD	Extensible framework with UI for visualizing and comparing evaluation results
AI Monitoring	Define events with natural language descriptions	Track issues, logical errors, and custom agent behaviors with semantic event definitions
SQL Access	Built-in SQL editor for all data	Query traces, metrics, and events; bulk create datasets from queries (also available via API)
Dashboards	Powerful dashboard builder	Full SQL query support for traces, metrics, and events
Data Annotation & Datasets	Custom data rendering UI	Fast annotation and dataset creation for evaluations

Quick Start

Self-Hosting with Docker Compose

Laminar is straightforward to self-host locally. Clone the repository and start services with Docker Compose:

Laminar 的本地自托管非常简便。克隆仓库后，使用 Docker Compose 启动服务：

git clone https://github.com/lmnr-ai/lmnr
cd lmnr
docker compose up -d

This launches a lightweight but fully functional version of the stack. Access the UI at http://localhost:5667. For production environments, we recommend using the managed platform or running docker compose -f docker-compose-full.yml up -d.

这将启动一个轻量级但功能完整的服务栈。在浏览器中访问 http://localhost:5667 即可使用 UI。生产环境建议使用托管平台或运行 docker compose -f docker-compose-full.yml up -d。

Enabling AI Monitoring (Signals)

To enable the AI monitoring feature in self-hosted mode, set the GOOGLE_GENERATIVE_AI_API_KEY environment variable in your .env file (required by both app-server and frontend):

要在自托管模式下启用 AI 监控（Signals）功能，请在 .env 文件中设置 GOOGLE_GENERATIVE_AI_API_KEY 环境变量（后端和前端均需要此配置）：

# In .env at the repo root
GOOGLE_GENERATIVE_AI_API_KEY=your_key_here

SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。 Integration Quickstart

TypeScript SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。

Install the SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。 and all instrumentation packages:

安装 SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。及所有检测工具包：

npm add @lmnr-ai/lmnr

Initialize tracing for LLM calls:

初始化 LLM 调用追踪：

import { Laminar } from '@lmnr-ai/lmnr';
Laminar.initialize({ projectApiKey: process.env.LMNR_PROJECT_API_KEY });

Wrap functions with the observe decorator to trace inputs and outputs:

使用 observe 包装函数以追踪输入和输出：

import { OpenAI } from 'openai';
import { observe } from '@lmnr-ai/lmnr';

const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

const poemWriter = observe({name: 'poemWriter'}, async (topic) => {
  const response = await client.chat.completions.create({
    model: "gpt-4o-mini",
    messages: [{ role: "user", content: `write a poem about ${topic}` }],
  });
  return response.choices[0].message.content;
});

Python SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。

Install the SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。 with all instrumentation:

安装 SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。及所有检测工具：

pip install --upgrade 'lmnr[all]'

Initialize and trace functions with the @observe() decorator:

初始化并使用 @observe() 装饰器追踪函数：

import os
from openai import OpenAI
from lmnr import observe, Laminar

Laminar.initialize(project_api_key="<LMNR_PROJECT_API_KEY>")
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

@observe()
def poem_writer(topic):
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": f"write a poem about {topic}"}],
    )
    return response.choices[0].message.content

Technical Architecture

Laminar is built for high performance, leveraging modern technologies to handle real-time observability at scale:

Laminar 基于高性能架构设计，利用现代技术应对大规模实时可观测性需求：


Component	Technology	Benefit
Core Engine	Rust一种注重安全、并发和性能的系统编程语言，通过所有权系统在编译时防止内存错误，适合高性能计算场景。 🦀	High memory safety and runtime performance
Trace Viewer	Custom realtime engine	View traces as they happen without lag
Full-Text Search	Ultra-fast engine over span data	Instant querying across large trace volumes
Data Export	gRPCGoogle开发的高性能远程过程调用框架，支持多种编程语言，常用于微服务通信，本文中提及的API服务器使用gRPC网关。 exporter	Efficient streaming of tracing data

Summary

Laminar provides a comprehensive, open-source observability solution specifically designed for AI agent workflows. With native OpenTelemetry一个开源的观测性框架，用于收集、处理和导出遥测数据（如指标、日志和追踪），RΞASON内置兼容此框架以实现可观测性。 support, powerful tracing capabilities, built-in evaluation tools, and a high-performance Rust一种注重安全、并发和性能的系统编程语言，通过所有权系统在编译时防止内存错误，适合高性能计算场景。 core, it addresses the unique monitoring needs of modern LLM-based applications. Whether self-hosted via Docker or used through the managed platform, Laminar offers a complete toolkit for understanding, debugging, and optimizing AI agent behavior from development through production.

Laminar 提供了一套全面且开源的可观测性解决方案，专为 AI 智能体工作流设计。凭借原生 OpenTelemetry一个开源的观测性框架，用于收集、处理和导出遥测数据（如指标、日志和追踪），RΞASON内置兼容此框架以实现可观测性。支持、强大的追踪能力、内置评估工具以及高性能的 Rust一种注重安全、并发和性能的系统编程语言，通过所有权系统在编译时防止内存错误，适合高性能计算场景。核心引擎，它能够满足现代基于大语言模型应用的独特监控需求。无论是通过 Docker 自托管，还是使用托管平台，Laminar 都提供了从开发到生产全流程理解、调试和优化 AI 智能体行为的完整工具集。

For complete documentation, visit docs.laminar.sh. Community discussions and support are available on Discord.

完整文档请访问 docs.laminar.sh。欢迎通过 Discord 参与社区讨论和获取支持。

常见问题（FAQ）

Laminar是什么？适用于哪些场景？

Laminar是开源LLM可观测性平台，专为AI智能体设计，提供全链路追踪、评估与监控，支持从开发到生产环境。

如何快速开始使用Laminar？

克隆GitHub仓库后运行docker compose up -d，访问http://localhost:5667。或通过npm/pip安装SDK，用@observe()装饰器一行代码自动追踪LLM调用。

Laminar支持哪些AI框架和语言？

支持Vercel AI SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。、LangChain、OpenAI、Anthropic等主流框架，提供TypeScript和Python SDK软件开发工具包，Laminar提供TypeScript和Python SDK用于代码插桩。，并兼容OpenTelemetry一个开源的观测性框架，用于收集、处理和导出遥测数据（如指标、日志和追踪），RΞASON内置兼容此框架以实现可观测性。标准。

AI Summary (BLUF)