portal-finops

Audit Google Cloud costs and enforce a $45 monthly budget for a financial data portal.

Updated Jul 27, 2026
One-click install
npx skills add https://github.com/ArthurZizumbo/karisma-data --skill portal-finops-arthurzizumbo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: portal-finops
Source: https://github.com/ArthurZizumbo/karisma-data/tree/main/.claude/skills/portal-finops
Command: npx skills add https://github.com/ArthurZizumbo/karisma-data --skill portal-finops-arthurzizumbo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Cloud projects silently drift over budget through forgotten Cloud Run instances, orphaned container images, and uncontrolled LLM token usage. This Skill gives you a repeatable audit routine to keep the Portal Centralizado de Datos Financieros under $45 USD/month with a billing alert at 50%. ## Core Features & Use Cases - Cost Auditing: Verify Cloud Run scale-to-zero, GCS 7-day lifecycle policies, Artifact Registry cleanup, and active billing budgets via gcloud and gsutil commands. - LLM Spend Tracking: Read real token consumption from OTel llm.usage.* span attributes instead of manual estimates, and confirm only Gemini Flash-Lite is used. - Savings Levers: Enforce tool-call budgets (max 5 per query), response caching, and SSE stream cancellation to stop unbilled token waste. - Use Case: Before a weekly demo, run the audit checklist to confirm min_instances = 0 on both Cloud Run services, the $45 budget alert is active, and projected LLM spend stays under $15/month. ## Quick Start Run a full cost audit of the portal and report whether Cloud Run, GCS lifecycle, billing alerts, and Gemini token usage are all within the $45 monthly budget.

Frequently Asked Questions about portal-finops

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

FAQPage Schema
How do I audit Google Cloud costs for a small project?

Run gcloud and gsutil checks for Cloud Run minScale settings, GCS bucket lifecycle rules, and active billing budgets. Combine these with a Terraform google_billing_budget resource that alerts at 50%, 90%, and 100% of your monthly limit.

How to track Gemini API token usage and cost?

Read the llm.usage.prompt_tokens, completion_tokens, and total_tokens attributes from OpenTelemetry llm.call spans. Multiply tokens by the Flash-Lite rate to get cost per day, and check llm.model to confirm no accidental model upgrades.

How do I set up a GCP billing budget alert with Terraform?

Define a google_billing_budget resource with your billing account, a specified_amount in USD, and threshold_rules for the alert percentages. This Skill uses a $45 budget with alerts at 50%, 90%, and 100%.

Does Cloud Run scale-to-zero reduce costs?

Yes, setting min_instances to 0 means Cloud Run charges nothing when idle outside business hours. The tradeoff is cold starts, which should be monitored in production smoke tests, with a documented temporary exception only for demo days.

Why is my LLM bill higher than expected?

Common causes are agent loops exhausting the tool-call budget, missing response caches, and streams that keep generating after the user stops. Check llm.tool_calls.count in traces, enable caching, and enforce real SSE cancellation to cut unbilled tokens.