A comprehensive skill for integrating Langfuse observability, prompt management, and evaluation into AI coding agent workflows.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install langfuse
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install langfuse using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Langfuse?
Langfuse is an open-source LLM engineering platform designed to handle tracing, prompt management, and evaluation for complex AI applications. This skill enables Openclaw Skills users to seamlessly connect their codebases with Langfuse, whether using the cloud version or a self-hosted instance. It prioritizes SDK-based integrations for Python and JavaScript/TypeScript to ensure robust observability and high-performance data handling.
By leveraging this skill, developers can move beyond hardcoded prompts and opaque agent executions. It facilitates deep visibility into LLM calls, costs, and latencies, while providing a structured way to manage prompt versions and collect evaluation scores. Whether you are auditing existing traces or building a brand-new evaluation pipeline, this tool ensures your Openclaw Skills setup is production-ready and data-driven.
Langfuse Use Cases
- Migrating hardcoded prompts into centralized Langfuse prompt management for better version control.
- Implementing observability and tracing in AI agents to debug execution steps and monitor performance.
- Collecting user feedback and automated evaluation scores to improve LLM output quality over time.
- Querying traces, metrics, and sessions to build custom analytics dashboards or debugging tools.
- Configuring self-hosted Langfuse environments with custom domains and private infrastructure.
How Langfuse Works
- Identification of the project environment, distinguishing between cloud-hosted and self-hosted Langfuse deployments.
- Configuration of authentication using public and secret keys alongside the appropriate base URL within the Openclaw Skills environment.
- Implementation of SDK-based instrumentation for Python or JS/TS to capture generations, spans, and traces.
- Centralization of prompts by replacing inline literals with dynamic fetch and render calls via the Langfuse API.
- Attachment of evaluation metrics and scores to specific observations or trace sessions for continuous improvement.
Langfuse Setup
To get started with this skill in your Openclaw Skills environment, configure the required environment variables:
export LANGFUSE_PUBLIC_KEY="pk-lf-..."
export LANGFUSE_SECRET_KEY="sk-lf-..."
export LANGFUSE_BASE_URL="https://cloud.langfuse.com" # Use your self-hosted URL if applicable
For Python projects, ensure you have the latest SDK installed:
pip install langfuse
For JS/TS projects:
npm install langfuse
Langfuse Data Schema & Taxonomy
The skill organizes LLM engineering data into a structured hierarchy to ensure clarity and searchability within Openclaw Skills.
| Entity | Description | Key Attributes |
|---|---|---|
| Traces | Top-level execution logs | trace_id, user_id, session_id, tags |
| Observations | Individual spans or generations | name, model, latency, tokens, cost |
| Prompts | Managed prompt templates | name, version, content, config |
| Scores | Evaluation and feedback data | name, value, type, comment |
| Datasets | Collections for repeatable testing | name, description, items |
Langfuse Advanced Features
- Full support for self-hosted deployments with custom URL routing and platform compatibility checks.
- High-performance asynchronous ingestion using the latest SDK namespaces such as api.observations and api.metrics.
- OpenTelemetry-compatible instrumentation for standardized observability across different application stacks.
- Dynamic prompt fetching with built-in caching, retries, and fallback mechanisms for high reliability in production.
- Multi-level scoring capabilities, allowing granular feedback to be attached to traces, generations, or dataset runs.
SKILL.md
Loading
Related Openclaw Skills
Code Debug
A high-precision diagnostic skill for identifying and repairing software bugs with minimal code changes.
Stock Auto Trade Skill
A comprehensive automation skill for executing stock trades, managing holdings, and querying accounts via a local HTTP API bridge.
Accessibility Checker
A professional tool for auditing and improving web accessibility to ensure WCAG compliance and inclusive user experiences.
Infoway Financial Data
A comprehensive MCP server providing real-time financial market data, sector analysis, and company fundamentals for over 20,000 instruments.
Code Orchestrator
An intelligent task router that coordinates specialized AI agents to handle complex coding lifecycles from exploration to deployment.
Code Architect
Code Architect generates minimal, actionable implementation plans and architectural designs optimized for your existing local codebase.
Featured*
