analyze-cloud-costs

Analyze Langfuse Cloud infrastructure costs using Metabase cost marts and dashboards.

34.0k|3.7k|Updated May 18, 2023
One-click install
npx skills add https://github.com/langfuse/langfuse --skill analyze-cloud-costs
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: analyze-cloud-costs
Source: https://github.com/langfuse/langfuse/tree/main/.agents/skills/analyze-cloud-costs
Command: npx skills add https://github.com/langfuse/langfuse --skill analyze-cloud-costs

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Understanding where Langfuse Cloud infrastructure spend goes requires querying multiple Metabase cost marts with correct table IDs, field IDs, and filters. This Skill provides the workflow, query patterns, and caveats needed to produce evidence-backed cost analysis without guessing at data structure.

Core Features & Use Cases

  • Provider and Service Breakdowns: Split costs by AWS versus ClickHouse, then drill into service, usage type, operation, and account using the unified cost mart.
  • Headline Daily Metrics: Retrieve total cost, per-provider cost, tracing event counts, and cost per 100k events from the daily totals mart.
  • Regression Analysis: Compare a recent complete-day window against a prior baseline to explain cost spikes or trends.
  • Use Case: When leadership asks why cloud spend jumped last week, query the daily totals mart, split by provider, then drill into the dominant service's usage types to name the exact cost driver.

Quick Start

Use the analyze-cloud-costs skill to explain the main Langfuse Cloud cost drivers over the past 30 days from Metabase.

Frequently Asked Questions about analyze-cloud-costs

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

FAQPage Schema
How do I analyze Langfuse Cloud infrastructure costs?

Query the Metabase cost marts starting with the daily totals table for headline metrics, then split by provider, service, and usage type in the unified cost mart. The skill's reference file lists the exact table IDs and field IDs to use.

How to break down AWS versus ClickHouse cloud spend?

Use the unified mart_daily_cost_chart table, aggregate sum of cost_usd, and group by service_provider. Then group by service_name within the dominant provider for a service-level split.

Why is current-day AWS cost data incomplete in Metabase?

AWS CUR rows can arrive late, so current-day AWS cost is not final. For stable analysis, filter to the last complete UTC days rather than including today.

What Metabase tables contain Langfuse Cloud cost data?

Four marts exist: mart_daily_cost_chart for unified provider breakdowns, mart_daily_cost_with_events for daily totals and cost per 100k events, mart_aws_cost_daily_by_service for AWS CUR detail, and mart_clickhouse_daily_cost for ClickHouse costs.

What are the limitations of Metabase cost mart analysis?

ClickHouse rows are labeled cost_usd but the source metric is ClickHouse credits, which matters for billing interpretation. Field IDs can also change if Metabase models are rebuilt, requiring a metadata lookup before rerunning queries.