What does this plugin do?

It captures traces, spans, and session telemetry from Claude Code and exports them to Langfuse, so you can monitor, debug, and evaluate what your agent does during each task.

Do I need a Langfuse account?

Yes. The plugin exports telemetry to Langfuse, so you need somewhere to send that data before traces will be available for review.

Can I use it for debugging agent behavior?

Yes. Reviewing exported traces and spans in Langfuse lets you see what the agent did during a task and debug problems.

langfuse-observability

PackMonitoring & ops

Langfuse observability is a plugin for Claude Code that captures traces, spans, and session telemetry from your agent's work and exports them to Langfuse. It gives you a record of what the agent did during each task, so you can review its behavior, debug problems, and run evaluations.

Unavailable. Delivery for this kind is on the roadmap — not serving yet.

Have a Claude Code setup and access to a Langfuse account or instance.

What your AI can do with it

Captures traces of agent activity during each task
Records spans within each trace
Collects session telemetry from Claude Code
Exports traces, spans, and telemetry to Langfuse
Supports LLM monitoring and debugging in Langfuse
Supports evaluation workflows using exported data

Getting started

Have a Claude Code setup and access to a Langfuse account or instance.
Install the langfuse-observability plugin.
Configure the plugin with your Langfuse connection details.
Run tasks in Claude Code and review the exported traces in Langfuse.

Signals

GitHub stars: 37k
Forks: 4k
Last commit: Sep 2026

Questions

What does this plugin do?: It captures traces, spans, and session telemetry from Claude Code and exports them to Langfuse, so you can monitor, debug, and evaluate what your agent does during each task.
Do I need a Langfuse account?: Yes. The plugin exports telemetry to Langfuse, so you need somewhere to send that data before traces will be available for review.
Can I use it for debugging agent behavior?: Yes. Reviewing exported traces and spans in Langfuse lets you see what the agent did during a task and debug problems.

Advanced

Item type: plugin
Key: anthropics-claude-plugins-official-langfuse-observability
Source: github.com/anthropics/claude-plugins-official

github.com/anthropics/claude-plugins-official