Claude Usage Companion

See Claude usage across claude.ai and Claude Code in one place, with history and forecasting.

Claude Usage Companion puts your Claude usage in one place: session and weekly limit bars, weeks of history as charts, a burn-rate forecast ("at this pace, you'll hit your limit around 8pm"), pay-as-you-go usage-credit spend, and configurable threshold alerts — all from data your own logged-in claude.ai session already receives, read locally in your browser. An optional, off-by-default warning tells you when you're on pace to run out before a limit resets. Install the optional local companion program with one command — npm install -g @rehberodhano/claude-usage-companion-daemon — and unlock Claude Code CLI attribution too: token totals by project and model, a rough estimate of how much of your weekly limit the CLI uses, full-text search across your own past sessions with a one-click resume command, and warnings before Claude Code's 30-day log cleanup deletes a session you might want to keep. The same companion program also powers Guardrails: see and override Claude Code's permission rules for a project across global/project/local scope, apply a set of recommended protections in one click, preview and edit its CLAUDE.md, see which subagent uses which model (and change it), and get a quick summary of what changed since you last looked. New Project (optional, needs the local helper): describe what you're building, and the Companion creates a ready-to-run starter with a Claude Code configuration, plus an optional check that it installs and its tests pass. Nothing here has an account, a server, or analytics. Everything is stored locally in your browser and, for the optional companion program, on your own machine only — see the privacy policy for exactly what's read and why every permission exists. This is an independent, unofficial project — not affiliated with or endorsed by Anthropic.