Instruments gateway functions with @observe() tracing for LLM applications.
npx skills add https://github.com/edupazogle/agentic_cockpit --skill langfuse-edupazogle
Please help me install this Agent Skill. Skill: langfuse Source: https://github.com/edupazogle/agentic_cockpit/tree/main/.agents/skills/langfuse Command: npx skills add https://github.com/edupazogle/agentic_cockpit --skill langfuse-edupazogle
Langfuse integration helps you reliably add observability to LLM-powered systems, manage prompts/datasets/evals, and query trace data without losing performance or leaking sensitive information.
Use the langfuse skill to instrument your gateway by asking the AI to provide a step-by-step plan for adding @observe() to your LLM/tool calls and propagating trace IDs into scenario_runs while ensuring no PII is included.
High-intent search queries and answers about installing and using this skill.
To add LLM tracing, you instrument gateway functions with the mandatory @observe() decorator, ensuring every LLM or tool-touching call creates trace spans and propagates Langfuse trace IDs for cockpit deep-links.
Langfuse integration enforces safe trace hygiene by applying defensive handling to prevent PII and sensitive credentials from being recorded in trace inputs, outputs, or span metadata during LLM observability.
Yes, you can manage prompt and evaluation assets and run dataset-driven evals by using langfuse-cli workflows to discover resources and query or modify traces, prompts, datasets, scores, and evals.
Self-hosted or private deployments require correct configuration of the LANGFUSE_HOST environment variable and valid API keys to connect your LLM application and enable end-to-end observability.
The best way to query trace data is using langfuse-cli operational workflows, which allow you to discover resources and query or modify traces, prompts, datasets, scores, and evals directly from your environment.
Propagating Langfuse trace IDs into scenario runs ensures end-to-end observability across your LLM application, linking individual evaluation executions to broader trace contexts for accurate cockpit deep-links.
jurislmManage Langfuse prompts, traces, observations, and scores via MCP.
ajbmachonAnalyze Langfuse traces, errors, costs, and evaluation results to diagnose AI issues.
MooseGoose0701Analyze agent execution traces to debug steps, LLM calls, and tool usage.
MapleTechLabsConfigure OpenTelemetry instrumentation for Next.js apps with Maple observability.
