skills.rest
HarmAalbers
HarmAalbersCOMMUNITY
Explore all skills by this author →
GitHub

langfuse

Query and manage a self-hosted Langfuse instance via CLI and documentation.

1|Updated Dec 11, 2025
Search Tags:#cli#prompt management#tracing#langfuse#self-hosted#eval#llm observability
One-click install
npx skills add https://github.com/HarmAalbers/claude-requirements-framework --skill langfuse-harmaalbers
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: langfuse
Source: https://github.com/HarmAalbers/claude-requirements-framework/tree/main/.agents/skills/langfuse
Command: npx skills add https://github.com/HarmAalbers/claude-requirements-framework --skill langfuse-harmaalbers

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires langfuse-cli, and includes references (resource) components.

What problem does it solve? Working with this project's self-hosted Langfuse v3 stack requires knowing the CLI, credential conventions, prompt sync workflow, and tracing setup — this Skill consolidates all of it so you can query traces, manage prompts, and debug observability without rediscovering project-specific rules. ## Core Features & Use Cases - Langfuse CLI data access: Discover API resources with langfuse-cli api __schema, then list or get traces, prompts, scores, sessions, and observations against the self-hosted instance at localhost:3000. - Project-specific observability guidance: Covers the V3 review tracing stack (observability.py, prompts.py, eval.py), the file-first prompt sync workflow, judge calibration, and error analysis of /v3-review traces. - Documentation retrieval: Fetch Langfuse docs via llms.txt, markdown page fetches, or the search-docs API, with caveats for the pinned self-hosted v3 deployment. - Use Case: You want to inspect why a /v3-review run scored low — source credentials from infra/.env, query the trace's GENERATION observations via the CLI, then follow the error-analysis reference to build an annotation queue. ## Quick Start Ask the assistant to list recent Langfuse traces from the self-hosted instance using the langfuse-cli with credentials sourced from infra/.env.

Frequently Asked Questions about langfuse

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query Langfuse traces from the command line?▼

Use npx langfuse-cli api to query traces without installing anything. Start with `langfuse-cli api __schema` to discover resources, source credentials from infra/.env, and export LANGFUSE_BASE_URL pointing to your instance before running list or get actions.

How do I sync prompts to the Langfuse prompt registry?▼

Edit the Jinja2 template in hooks/lib/llm/prompts/, then run scripts/sync_prompts_to_langfuse.py to push it with the production label. Files are the source of truth — never edit prompts in the Langfuse UI because the next sync overwrites them.

Why is my Langfuse CLI connection failing?▼

Connection failures usually mean the docker stack is down or credentials are missing. Check `docker compose -f infra/docker-compose.yml ps` first, then verify LANGFUSE_PUBLIC_KEY and LANGFUSE_SECRET_KEY exist in infra/.env without printing their values.

Does this work with Langfuse Cloud or only self-hosted?▼

This Skill is tailored to a self-hosted Langfuse v3 deployment at localhost:3000 with a pinned docker-compose stack and Python SDK v3. Documentation fetched from langfuse.com defaults to Cloud and current SDK versions, so cross-check snippets against the pinned versions.

Why are trace inputs and outputs null in my Langfuse traces?▼

In OpenTelemetry-instrumented apps like this project's /v3-review, trace-level input/output is often null because content lives on the GENERATION observation. Target annotation queues at objectType OBSERVATION with the generation observation ID instead of TRACE.

When should I not use the Langfuse CLI for score creation?▼

The langfuse-cli score-create wrapper cannot pass the required score value argument in version 0.0.10. Use SDK experiment evaluators or the REST API POST /api/public/scores endpoint instead for creating scores programmatically.

Related Skills

Explore More Software Engineering Skills→
AuraHQ-aiAuraHQ-ai
1

langfuse

Query and manage Langfuse traces, prompts, datasets, scores, and sessions via CLI.

Official
Intermediate
EntelligentsiaEntelligentsia
1

evaluation-harness

Evaluate AI model outputs against predefined criteria using golden datasets and LLM-as-judge.

Official
Advanced
grafanagrafana
213

app-observability

Generate RED metrics, RUM data, and LLM telemetry for Grafana Cloud.

Official
Advanced
hiddink-aihiddink-ai
1

professor-triage

Analyze GitHub issues against the codebase for automated triage and priority assessment.

Official
Advanced