Orpheus
Guides

Glossary

The terms behind Claude usage limits, coding agents and token adoption, each in a sentence or two. Updated .

A

Account

One Claude subscription (Pro, Max or Ultra) signed in on a machine. An Orpheus plan sets how many accounts you can connect: 1, 2, 4, 8 or 12.

See Running two Claude accounts on one machine.

Agent view

Claude Code's own screen, opened with claude agents, that lists the background sessions on one machine. A research preview as of November 2026.

See Dashboards for Claude Code agents.

Ask

A question an agent sends you through Orpheus when it needs a decision. You answer on any device and the answer is typed back into the agent's session.

Auto-move

An Orpheus setting that moves idle agents off an account that reached its threshold and onto the connected account with the most room. An agent in the middle of a turn is left alone until it goes idle. Needs at least two accounts.

See also Threshold, Idle session.

C

Claude Code

Anthropic's coding agent. It runs in a terminal (and on the web and in the Claude app) and uses the Claude subscription or API key you sign in with.

CLAUDE_CONFIG_DIR

The environment variable that tells Claude Code which configuration folder to use instead of the default ~/.claude. Anthropic's docs describe it as useful for running multiple accounts side by side.

See also Config folder.

Codex

OpenAI's coding agent CLI. On ChatGPT plans it has its own usage limits, measured per 5-hour period, separate from Claude limits.

Config folder

The folder where Claude Code keeps one account's sign-in, settings and session history. The default is ~/.claude; each extra account gets its own folder.

See Managing multiple accounts.

Connector

The small program you install on a machine for Orpheus. It dials out over one secure WebSocket, reads the tmux sessions your policy file exposes and types into them. It has no way to run commands.

See Security.

F

5-hour limit

Also called the session limit. A cap on Claude usage inside a five-hour window that starts with your first message and resets in full five hours later. Ultra has no 5-hour limit for verified accounts.

See When the 5-hour limit resets.

G

Git worktree

A second working folder for the same git repository, on its own branch. Running one agent per worktree keeps parallel agents from editing the same files.

See Running agents in parallel.

I

Idle session

An agent session that is waiting for input and not in the middle of a turn. Only idle sessions are moved between accounts.

M

Max

The $100 a month Claude plan, called Max 5x until October 30, 2026. It gives more usage than Pro every 5 hours and every week.

See Which Claude plan for Claude Code.

Model-specific limit

A weekly cap on one model family, shown in Claude Code as an Opus or Sonnet limit. Switching to a model outside that family keeps you working; the session and weekly limits still apply.

Move to

An Orpheus action on every session that lists your other accounts with the room each has left, so you can move that agent yourself.

P

Policy file

The file ~/.config/orpheus-connector/policy.json on your machine. It decides which sessions Orpheus sees, whether input is allowed, whether transcripts are shared and whether usage is read. Orpheus cannot change it.

See Setup guide.

Pro

The $20 a month Claude plan ($17 a month billed yearly). It includes Claude Code with the smallest usage limits of the paid plans.

R

Remote Control

Anthropic's feature for driving a Claude Code session running on your own machine from claude.ai or the Claude app. The machine has to stay on.

See Controlling Claude Code from your phone.

S

Statusline

A script Claude Code runs to draw its bottom status bar. For Pro, Max and Ultra logins it receives the 5-hour and 7-day usage percentages and reset times as JSON.

See Checking your Claude Code usage.

T

Team seat

A seat on the Claude Team plan (2 to 150 seats). Standard seats cost $25 a month ($20 billed yearly); premium seats cost $125 ($100 billed yearly) with five times the usage of a standard seat.

See Rolling out Claude Code to a team.

Threshold

The share of an account's 5-hour or weekly limit at which Auto-move starts moving idle agents off it. You set it per account.

tmux

A terminal multiplexer: it keeps terminal sessions running after you disconnect and lets other programs read and type into them. Orpheus sees agents that run inside tmux.

See Running Claude Code in tmux.

Token

The unit AI models count text in. Subscriptions turn token use into percentages of a 5-hour and a weekly limit; the API bills tokens directly.

Token adoption

How widely and how usefully the people in a team or company use AI models, often measured through token use, active users and the share of paid capacity used. Leaderboards of raw token counts have been gamed and withdrawn at some companies.

See What token adoption is.

Tokenmaxxing

Slang for using as many AI tokens as possible, often to climb an internal usage leaderboard. It counts volume, not results.

See Should you rank engineers by token use.

U

Ultra

The $200 a month Claude plan, called Max 20x until October 30, 2026. Anthropic's pricing page lists no 5-hour limit for verified accounts.

Usage credits

Anthropic's pay-as-you-go extra usage for paid Claude plans, billed at standard API rates. When you reach a limit, requests can continue on credits up to a spend cap you set.

See How usage credits work.

Utilization

The share of an account's weekly limit actually used by the time it resets. A simple way to see whether a subscription is earning its price.

See Using the capacity you pay for.

W

Weekly limit

A cap on Claude usage over seven days, across all models, with a fixed reset time per account shown in Settings, Usage. Anthropic added weekly limits for Pro and Max on August 28, 2025.

See How the weekly limit works.