instrument-llm-analytics

What problem does it solve?

Instrumentation of LLM calls to automatically generate and push $ai_generation events to PostHog, enabling observability of model usage, latency, tokens, and costs across your stack.

Core Features & Use Cases

• Auto-instrument LLM calls across providers and frameworks using OpenTelemetry and PostHog instrumentation to generate $ai_generation events with model, provider, input/output tokens, latency, and cost data.
• Link events to users and traces, enabling per-user analytics, traces, spans, and sessions for debugging and performance optimization.
• Quick start guidance for installation, environment configuration, and minimal code changes to instrument LLM calls without restructuring your architecture.
• Supports multiple providers and frameworks as listed in the SKILL.md to ensure broad coverage.

Quick Start

Configure your project to enable PostHog LLM analytics instrumentation, install the required SDKs, and instrument your LLM calls with minimal code changes.