What is MCP Tool Search Auto Mode?

Introduced in version 2.1.7, it automatically defers loading MCP tool descriptions if they take up more than 10% of the context, saving significant space and cost.

How does selective file reading reduce costs?

By using specific @ file mentions or line ranges instead of reading entire directories, you significantly reduce the input tokens Claude must process, lowering the session cost.

When should I use /compact versus /clear?

Use /compact when context is over 70% full but you want to continue the same task. Use /clear when switching to an unrelated task or when you want to reset costs to zero.

How can I check my current session costs in Claude Code?

You can use the /cost command to view your current session's input and output token counts, the estimated cost, and the percentage of context window used.

Which Claude model is best for cost-effective coding?

Claude Haiku is approximately 15x cheaper than Opus and is ideal for simple tasks, file exploration, and basic edits. Sonnet is the balanced default for general development.

Claude Cost Optimization

Name: Claude Cost Optimization
Author: hgeldenhuys

byhgeldenhuys

•

生産性とワークフロー

Minimizes Claude Code API expenditures through strategic token management, model selection, and usage monitoring.

概要

The Cost Optimization skill provides a comprehensive framework for managing and reducing API spend while using Claude Code. It empowers developers to monitor real-time token usage via the /cost command, analyze historical spending with /stats, and implement high-impact reduction strategies like context compacting and selective file reading. By matching tasks to the most cost-effective models—Haiku, Sonnet, or Opus—and leveraging automatic MCP tool optimization, this skill ensures that development workflows remain efficient and budget-friendly without sacrificing output quality.

主な機能

1 GitHub stars
Real-time session cost and token monitoring via the /cost command
Historical usage analysis and model distribution tracking with /stats
Strategic context management using /compact and /clear operations
Model selection guidance to balance performance and price for specific tasks
Automated MCP tool deferral to preserve context window space

ユースケース

Prototyping and file exploration using low-cost models like Claude Haiku
Managing large-scale refactors where context window growth can lead to high costs
Auditing and controlling developer API budgets through session tracking

概要

主な機能

1 GitHub stars
Real-time session cost and token monitoring via the /cost command
Historical usage analysis and model distribution tracking with /stats
Strategic context management using /compact and /clear operations
Model selection guidance to balance performance and price for specific tasks
Automated MCP tool deferral to preserve context window space

ユースケース

Prototyping and file exploration using low-cost models like Claude Haiku
Managing large-scale refactors where context window growth can lead to high costs
Auditing and controlling developer API budgets through session tracking