Langfuse Agent Skills - Setup Guide¶
Source: langfuse/skills GitHub: langfuse/skills Skills: 6 skills · 13.1K total installs Category: LLM Observability First Seen: catalogued August 15, 2026 evening sweep Quality Tier: 🟢 Production (official org - Langfuse, the LLM observability platform)
Langfuse's official skills teach agents to instrument LLM calls with tracing, manage prompts, and run observability workflows. For a multi-model operation like CorpusIQ - DeepSeek primary, Claude fallbacks, cost tracking - tracing is how you see where tokens and money actually go.
Installation¶
npx skills add langfuse/skills
Prerequisites¶
| Requirement | Details |
|---|---|
| Node.js + npx | For the skill installer |
| Langfuse account | Self-hosted or cloud, with API keys |
| Langfuse SDK | In the instrumented application |
What It Provides¶
| Skill | Installs | Purpose |
|---|---|---|
| langfuse | 12.7K | Core tracing and observability |
| langfuse-observability | 217 | Observability patterns |
| langfuse-prompt-migration | 87 | Migrate prompts into Langfuse management |
| langfuse-api | 56 | API reference workflows |
| langfuse-cli | 1 | CLI usage |
| skill-creator | 4 | Skill authoring helper |
Quick Start¶
- Install:
npx skills add langfuse/skills - Start with the core
langfuseskill - it carries 97% of the suite's installs - Ask: "instrument my agent loop with Langfuse tracing and show cost per model"
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Model cost tracking | Trace every LLM call to attribute spend per model and task |
| Agent debugging | Full traces of multi-step agent runs instead of guessing where a loop degraded |
| Prompt management | prompt-migration to move prompts out of code and into versioned management |
| Client observability | Reference architecture for clients building LLM-powered products |
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations / Verification¶
- Publisher-page install counts verified; individual skill audit pages were not fetched for this multi-skill suite
- The core skill dominates: observability, API, and CLI skills show sub-1K installs - treat as early content
- Requires a running Langfuse instance and SDK integration to be useful
npx skills add langfuse/skills # verify install works
Related¶
Powered by CorpusIQ