obs-cost

Monitor AI coding assistant costs and token usage across providers.

70|6|Updated Feb 23, 2026

Search Tags:#observability #token usage #grafana #cost tracking #promql #ai spending

Ready to run

npx skills add https://github.com/shepard-system/shepard-obs-stack --skill obs-cost

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: obs-cost
Source: https://github.com/shepard-system/shepard-obs-stack/tree/main/.claude/skills/obs-cost
Command: npx skills add https://github.com/shepard-system/shepard-obs-stack --skill obs-cost

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill provides detailed insights into the cost and token consumption of AI coding assistants, helping users understand their spending and resource utilization.

Core Features & Use Cases

Cost Monitoring: Track total spending and breakdown by AI model (e.g., Claude, Gemini, Codex).
Token Usage Analysis: Monitor input, output, and cache token counts per provider.
Session Tracking: Report on the number of sessions initiated per AI provider.
Use Case: Understand how much you've spent on Claude Code today, broken down by model, and compare it to the token usage of Gemini CLI.

Quick Start

Show me the cost and token usage report for the last 24 hours.

Frequently Asked Questions about obs-cost

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

FAQPage Schema

How do I track AI coding spend and token usage across multiple providers?▼

Track AI coding spend and token usage by analyzing Prometheus metrics scraped from AI tool usage. This process monitors total spending, input/output/cache token counts, and active sessions across providers like Claude, Gemini, and Codex.

Can I monitor cost breakdowns by AI model using PromQL and Grafana?▼

Yes, you can monitor cost breakdowns by AI model using PromQL queries against scraped metrics. This allows you to analyze spending by model and visualize the token consumption data within Grafana dashboards.

Do I need an API endpoint to query AI spending data?▼

Yes, an API endpoint is required to query the AI spending data. Alongside Prometheus metrics scraped from your AI tool usage, this endpoint enables the system to retrieve and report on detailed cost and token consumption.

What's the best way to compare token usage between Claude Code and Gemini CLI?▼

The best way to compare token usage between Claude Code and Gemini CLI is by analyzing their respective input, output, and cache token counts. This Skill evaluates session tracking and spending per provider to give you a direct comparison.

Why does monitoring observability for AI spending require Prometheus metrics?▼

Monitoring observability for AI spending requires Prometheus metrics because the analysis relies on data scraped directly from AI tool usage. This raw metric data is necessary to calculate costs, track tokens, and count active sessions accurately.

Does obs-cost work with session tracking for multiple AI coding assistants?▼

Obs-cost works with session tracking by reporting the number of sessions initiated per AI provider. It aggregates this session data with cost and token metrics to provide a comprehensive view of your AI coding assistant usage.