Every guide in one place

How to reduce Claude Code costs

Five ways to cut Claude Code token usage: compress context, quiet terminal output, stop re-reading the repo, make the agent write less, and watch your windows.

Claude Code usage limits: the 5-hour window, the weekly cap, and what each plan gets

How Claude Code usage limits work: the 5-hour window, the weekly cap, what Pro and Max get, what counts toward usage, and how to check what is left with /usage.

Why is Claude Code so expensive?

Four reasons Claude Code burns through tokens fast: verbose tool output, repeated context, multi-step debugging, and large codebase reads. Plus how to fix each.

Reduce Claude API costs

Practical levers for cutting Claude API spend: prompt caching, model tier selection, output length, and batch API. Plus the Claude Code-specific shortcut.

Codex is now part of ChatGPT: what changed and what still works

OpenAI merged the Codex app into ChatGPT on July 9, 2026. What was renamed, what kept the Codex name (the CLI, ~/.codex), and what the merger broke.

How to reduce Codex costs

Five ways to cut Codex token usage in the ChatGPT app and CLI: compress context, quiet terminal output, stop re-reading the repo, make the agent write less.

Codex usage limits in 2026: the 5-hour window, the weekly cap, and what each plan gets

How Codex usage limits work in the CLI and ChatGPT app: the 5-hour window (back since July 30), the weekly cap, plan allowances, and how to check what's left.

Why is Codex so expensive?

Four reasons OpenAI Codex burns through usage fast: verbose tool output, repeated context, multi-step debugging, and large codebase reads. Plus how to fix each.

What is Headroom?

Headroom is a desktop app built on the open-source Headroom CLI (headroom-ai). It cuts what Claude Code and Codex send by typically 25-50% a session.

How to install Headroom for Claude Code and ChatGPT Codex

Install Headroom on macOS, Windows, or Linux with no Python or pip, or run the free open-source CLI with headroom wrap claude or codex. Steps and which to pick.

Is Headroom a Claude Code plugin or skill?

The Headroom app is a local proxy, not a Claude Code plugin. The open-source CLI's plugin only starts that proxy, and the skill you may be after is Ponytail.

Headroom vs Ponytail vs RTK: what's the difference?

Headroom, Ponytail, and RTK aren't competitors; they cut token cost on different sides of the bill and run together. What each does and when to use it.

Headroom vs RTK: what's the difference?

RTK rewrites noisy terminal commands into token-lean equivalents. Headroom compresses everything else before it reaches the model. Why you run both.

Headroom vs Caveman: what's the difference?

Both compress agent context reversibly via a local proxy. The real split: a finished desktop app for Claude Code and ChatGPT Codex vs. an open-source toolkit.

Caveman alternatives: what to use instead

Looking for a Caveman alternative to cut Claude Code and ChatGPT Codex token usage? Headroom is the closest match, plus the free and DIY options worth knowing.

Headroom for desktop vs Headroom CLI: what's the difference?

Two tools share the name Headroom: the open-source CLI that compresses agent context, and the desktop app built on it. How they relate and which to pick.

Headroom for OpenCode: how to set it up

Connect OpenCode to Headroom to compress its context: the desktop connector, what it writes to opencode.json, the transport plugin, and headroom wrap opencode.

Changelog

A running log of what's shipped in Headroom: connectors, add-ons, savings layers, and platform support, newest first.

How to uninstall Headroom

Use Settings > Uninstall Headroom to remove managed routing, add-ons, credentials, and local data. Includes platform details and recovery steps.

Ponytail for Claude Code and Codex: how to install and use it

Ponytail makes Claude Code and Codex write the least code that works. The exact install commands for both, how to check it is on, and its levels and commands.

Benchmarks · October 6, 2026

Extra Headroom in production: Input -34%, Output -33% across 183 Claude Code users

Production data from 183 Claude Code users over one week: Extra Headroom removed 3.50B input tokens and avoided 132M output tokens, $27,268 at list prices.

Sustainability · September 24, 2026

How Headroom makes AI coding more sustainable

AI coding agents are energy hungry. Headroom kept 4.76 billion tokens a week out of data centers, about 19-192 kWh a year per user. Method and sources shown.

Measurement · August 17, 2026

How to present token savings percentages honestly

One day of real agent traffic, scored three ways: 15%, 38%, or 87%. Which savings percentage means anything, and why cache reads stay out of the denominator.

Usage limits · July 6, 2026

Hit the Claude Code weekly limit by Wednesday? Here's what's eating it

Why heavy Claude Code users burn the weekly limit in two or three days (sub-agents, tool output, re-read files) and what to do this week.

Claude Code · July 2, 2026

Sub-agents are why your Claude Code weekly limit vanished

Sub-agents spin up fresh contexts that don't inherit the prompt cache, so multi-agent runs re-buy context. How the tax works and how to shrink it.

Codex · June 15, 2026

Codex now meters by tokens: why your ChatGPT plan drains faster

Since April 2026, ChatGPT Plus, Pro, and Business meter Codex by tokens instead of messages. What changed, who it hurts, and how to adapt without upgrading.