langfuse

name

langfuse

description

General-purpose Langfuse integration for observability, evaluations, datasets, dataset items, experiment runs, remote experiment webhooks, dashboard widgets, dashboards, metrics, traces, scores, models, and filter discovery. Use for any Langfuse task including discovering project data, querying metrics, creating or managing datasets, running or comparing experiments, creating or managing LLM-as-a-Judge evaluators, and creating or arranging dashboard visualizations.

Langfuse

Use this as the single public Langfuse skill. The previous task-specific skills are preserved as internal sub-skills under references/sub-skills/.

Start

1. Read ../../AGENTS.md for shared Langfuse data model, credentials, storage, and API/DB access strategy.
2. Classify the user request with the routing map below.
3. Read each matching references/sub-skills/<name>/SKILL.md before acting.
4. Resolve any sub-skill references/... links relative to that sub-skill directory, not relative to this file.
5. When a sub-skill says to invoke, use, or delegate to another former skill, read the matching internal sub-skill instead of looking for a separate exposed skill.
6. Load only the sub-skills needed for the task. Do not read every internal sub-skill by default.

For destructive operations, direct database writes, status changes, or webhook configuration, follow the confirmation and safety checks in the selected sub-skill before applying changes.

Routing Map

User intent	Internal sub-skills to read
Discover traces, trace names, tags, environments, sessions, or users	discover-traces
Discover scores, score names, score types, or score sources	discover-scores
Discover datasets, dataset metadata, items, or runs	discover-datasets
Discover dataset runs	list-dataset-runs
Discover model definitions or pricing	discover-models
Discover trace filter options for evaluators	discover-filter-options
Query aggregate analytics or the metrics API	query-metrics
Create a dataset	create-dataset
Design dataset input or expected output schemas	design-dataset-schema
Add, update, archive, delete, list, or bulk import dataset items	manage-dataset-items
Trigger an experiment or dataset run	trigger-experiment
Configure remote experiment webhooks or payloads	configure-remote-experiment
Analyze experiment results or inspect per-item scores	analyze-experiment-results
Compare experiment runs or detect regressions	compare-experiments
Create an LLM-as-a-Judge evaluator	create-evaluator
List evaluator configurations	list-evaluators
Inspect evaluator prompts, versions, or job configs	inspect-evaluator
Update evaluator prompts, filters, mappings, model, or sampling	update-evaluator
Activate, deactivate, pause, resume, enable, or disable evaluators	toggle-evaluator-status
Delete or remove evaluators	delete-evaluator
List dashboard widgets	list-widgets
Suggest dashboard visualizations	suggest-widgets
Create a dashboard widget	create-widget
Update a dashboard widget	update-widget
Delete a dashboard widget	delete-widget
Create, update, delete, or arrange dashboards	manage-dashboard
Calculate dashboard widget placement	layout-widgets

Common Chains

• For dashboard creation from a vague request, read suggest-widgets, then create-widget, layout-widgets, and manage-dashboard as needed.
• For evaluator setup, read discover-traces, discover-filter-options, discover-models, and create-evaluator.
• For experiment setup, read discover-datasets, create-dataset, manage-dataset-items, configure-remote-experiment, and trigger-experiment as needed.
• For experiment review, read list-dataset-runs, analyze-experiment-results, and compare-experiments as needed.