Glossary
Glossary
Short, sourced definitions of the terms you meet when working with Claude Code and Codex, and how each shows up in Pairbar.
5
- 5-hour usage window
- The 5-hour usage window is Claude's session limit: a five-hour period that caps how much you can use Claude, including Claude Code, until it resets.
A
- Agent action
- An agent action is a request from an AI coding agent for a tool to do something on its behalf, such as starting another chat or running a job, rather than only reading information.
- Agent hooks
- Agent hooks are commands that an AI coding agent such as Claude Code or Codex runs automatically at set points in its lifecycle, for example when a session starts, before or after a tool call, on a permission request, or when a turn stops.
- Agent profile
- An agent profile is a named preset for the chats an AI agent asks to start, setting which app and model run it, how much effort it spends, what it may do, and when it's the right choice.
- Approval ticket
- An approval ticket is the record created for one request an AI agent makes, which waits for a decision, then follows the work it starts and holds the result for the agent to collect.
B
- Basket
- The basket is a Pairbar window layout that docks the official Claude and Codex desktop apps together with one Pairbar sidebar in a single frame, so the two behave like one window you move, resize and rearrange together.
C
- Claude Code agent
- A Claude Code agent, or custom subagent, is a Markdown file with YAML frontmatter that defines a reusable specialist, such as a code reviewer, that Claude Code can hand work to.
- Claude Code cloud session
- A Claude Code cloud session is a Claude Code session that runs on cloud infrastructure, by default an isolated virtual machine managed by Anthropic, instead of on your own computer.
- Claude Code session
- A Claude Code session is one saved conversation with Claude Code, tied to a project folder and stored on your machine so you can pick it up again later.
- Codex cloud environment
- A Codex cloud environment is the saved configuration Codex uses to build the container a cloud task runs in: which repository, which image, which setup commands, and what variables and internet access the agent gets.
- Codex cloud task
- A Codex cloud task is a piece of work you give Codex to run in an OpenAI-hosted container built from one of your cloud environments, instead of on your own machine.
- Codex sandbox modes
- Codex sandbox modes are the settings that limit what the commands Codex runs can read, write and reach on your machine: `read-only`, `workspace-write` or `danger-full-access`.
- Codex thread
- A Codex thread is a single conversation between you and OpenAI's Codex agent, made up of turns: each turn is one request plus the work the agent does for it.
F
- Fork (of a conversation)
- A fork of a conversation is a new chat that starts with a copy of an existing chat's history, up to its latest message or an earlier one, so you can try another direction without changing the original.
G
- Git worktree
- A git worktree is an additional working directory attached to the same Git repository, with its own checked-out branch and files but a shared history, created with `git worktree add`.
H
- Handoff
- A handoff is passing an unfinished task from one AI coding agent or chat to another along with the context the new one needs to carry on: what was asked, what was done, and the current state of the code.
- Headless mode
- Headless mode, also called non-interactive mode, is running an AI coding agent from a single command or script, so it works through a prompt and prints the result without an interactive interface.
M
- MCP server
- An MCP server is a program that exposes tools, resources and prompts to AI applications over the Model Context Protocol, either as a local process speaking over standard input and output or as a remote service over HTTP.
- Model Context Protocol (MCP)
- The Model Context Protocol (MCP) is an open standard for connecting AI applications to external systems such as data sources, tools and workflows.
O
- Overlay sidebar
- An overlay sidebar is a separate app that draws its own panel over part of another app's window, usually its sidebar, to add features while the original app keeps running unchanged underneath.
P
- Permission mode
- A permission mode is the Claude Code setting that decides which actions Claude may take in a session without asking you first.
- Plan usage limits
- Plan usage limits are the caps a Claude or ChatGPT subscription puts on how much you can use Claude Code or Codex within set time windows, usually a rolling five-hour window plus a weekly one.
- Push delivery
- Push delivery is sending the result of work an AI agent asked for back into that agent's chat as a new message, once the chat can take one, instead of waiting for the agent to collect it.
S
- Skill drift
- Skill drift is when copies of the same agent skill in different folders stop matching, so Claude Code, Codex or different projects load different versions of its instructions.
- Subagent
- A subagent is a helper agent that a main AI coding agent starts to handle one piece of work in its own separate context, handing back only the result.
V
- Voice dictation (for coding agents)
- Voice dictation is speaking text instead of typing it, with speech recognition turning your words into text in the field you're typing in; for coding agents such as Claude Code or Codex, it means speaking a prompt into the agent's message box.
W
- Workflow wave
- A workflow wave is a group of steps in a multi-step workflow that wait for the same earlier steps, so they can run side by side instead of one after another.
- Worktree task
- A worktree task is one piece of coding work done on its own branch in a separate git worktree, so one or more AI coding agents can work on it without touching your main checkout or each other's changes.
Try Pairbar free for 7 days
One sidebar for the Claude and Codex desktop apps. Everything on for 7 days, no card needed; then $49/year ($39 first year with LAUNCH), $9/month or $129 lifetime (first 200 buyers). macOS 26 or later, Apple silicon.