LMNR

Active
GitHub TypeScript Apache-2.0

Description

LMNR is an open-source observability platform for LLM and agent applications, focused on tracing, quality analysis, and production diagnostics.

Key Features

  • OpenTelemetry-native tracing SDK - One-line automatic tracing for Vercel AI SDK, Browser Use, LangChain, OpenAI, Anthropic and more
  • Extensible evaluation system - Unopinionated SDK and CLI for running evals locally or in CI/CD with UI visualization
  • AI monitoring signals - Define events with natural language to track issues, logical errors, and custom agent behavior
  • Built-in SQL editor - Query traces, metrics, and events directly with SQL, bulk create datasets from queries
  • Custom dashboard builder - Powerful builder for traces, metrics, events with custom SQL query support
  • Ultra-high performance - Written in Rust with realtime trace viewing, ultra-fast full-text search, and gRPC exporter

Use Cases

💡 Continuously monitor LLM and agent application health and performance in production
💡 Define monitoring rules in natural language to automatically detect agent behavior anomalies
💡 Use the SQL editor to deep-dive into trace data for performance bottleneck identification
💡 Run agent evaluations in CI/CD pipelines for continuous quality regression detection

Tags

observability tracing diagnostics python

Categories

📊 Observability

Quick Start

Visit laminar.sh to create a project and generate an API key. Python: pip install 'lmnr[all]' then Laminar.initialize(project_api_key='...'). TypeScript: npm add @lmnr-ai/lmnr then Laminar.initialize(). Wrap functions with @observe() decorator or observe(). Self-host: git clone then docker compose up -d.
Visit GitHub Visit Website View Docs

Related Projects