# AI coding agent glossary: ADE, worktrees, squads and more > The Tallos glossary defines the terms used in AI-assisted and multi-agent software development — from ADE, AI coding agent and git worktree to squads, best-of-N, MCP and vibe coding. Each entry answers “what is it?” in one to three plain sentences and links to the page that explains it in depth. - Canonical: https://runtallos.com/glossary - Português: https://runtallos.com/pt/glossario.md - Updated: 2026-09-29 ## Key facts - 26 terms, one plain definition each - Covers agents, workspaces, squads and review - Every entry links to a deeper guide - Also available in Brazilian Portuguese ## What is an ADE? An **ADE (agentic development environment)** is a desktop environment built for directing AI coding agents rather than typing every line yourself. It gives each task an isolated workspace, runs several agent CLIs side by side, shows which ones need you and puts diff review before merge. Read [what an ADE is](https://runtallos.com/what-is-an-ade). ## What is an AI coding agent? An **AI coding agent** is a program that uses a language model to complete a software task by acting on your project: it reads files, edits them, runs commands and tests, and repeats until the goal is met. Claude Code, Codex and Gemini CLI are common examples. Read [what AI coding agents are](https://runtallos.com/what-are-ai-coding-agents). ## What is an agent CLI? An **agent CLI** is an AI coding agent that runs as a command-line program in your terminal, such as `claude`, `codex`, `gemini` or `opencode`. Each vendor ships its own install command and sign-in. Tallos can install and sign in to supported agent CLIs from its [Connections page](https://runtallos.com/features/agent-connections). ## What is multi-agent development? **Multi-agent development** is building software with several AI coding agents working at the same time, each with a defined job and its own workspace, while a person or a leader agent coordinates and reviews the results. Read the [multi-agent development guide](https://runtallos.com/multi-agent-development). ## What are parallel agents? **Parallel agents** are two or more AI coding agents working on the same repository at the same moment, each on its own task or its own attempt at the same task, each in a separate checkout. See [how to run AI agents in parallel](https://runtallos.com/run-ai-agents-in-parallel). ## What is an isolated workspace? An **isolated workspace** is a separate copy of a project, with its own branch, files, terminals and browser tabs, where one task runs without touching any other task. In Tallos every workspace is a real git worktree. See [parallel workspaces](https://runtallos.com/features/parallel-workspaces). ## What is a git worktree? A **git worktree** is an extra working directory attached to the same git repository, checked out on its own branch. Several worktrees share one history, so they are a light way to give each AI agent its own files. Read [git worktrees for AI agents](https://runtallos.com/learn/git-worktrees-for-ai-agents). ## What is a squad? A **squad** is a Tallos team made of one leader agent and one or more member agents, with limits you set and a final review where you accept or discard the result. Templates include Build + review, 3 attempts and pick the best, and Leader divides the work. See [Squads](https://runtallos.com/features/squads). ## What is a leader agent? A **leader agent** is the agent in a squad that plans the work, hands out tasks to members, checks their results and reports back to you. It asks you a question only when it is truly blocked, and it cannot exceed the squad's round and member limits. See [Squads](https://runtallos.com/features/squads). ## What is a squad member? A **squad member** is an agent in a squad with a job you describe, such as “writes the tests,” plus its own agent, model and effort level. Each member works in its own worktree, and you can run several copies of the same member. Build one in [your first squad](https://runtallos.com/learn/your-first-squad). ## What is a round in a squad? A **round** is one cycle of a squad's work: the leader plans, dispatches members and reviews what comes back. Squads run at most 5 rounds by default, up to a hard maximum of 20, and the leader stops and reports when the limit is reached. See [Squads](https://runtallos.com/features/squads). ## What is best-of-N? **Best-of-N** is a pattern where the same objective goes to several agents that work independently, and the best result is kept. In Tallos it is the squad template “3 attempts, pick the best,” where the leader compares the attempts and recommends one. See the [best-of-three squad](https://runtallos.com/squads/best-of-three). ## What is build + review? **Build + review** is a pattern where one agent does the work and another reviews it, and they go back and forth until the reviewer approves. It is a Tallos squad template for refactors, migrations and risky changes. See the [build + review squad](https://runtallos.com/squads/build-and-review). ## What is agent orchestration? **Agent orchestration** is coordinating several AI agents so their work fits together: splitting an objective into tasks, dispatching them, tracking completion and collecting results. In Tallos, squads run on an orchestration layer that agents drive through the `tallos` CLI. Read about [multi-agent development](https://runtallos.com/multi-agent-development). ## What is native chat? **Native chat** is Tallos's readable chat view over a supported agent's terminal session, showing messages, tool calls and questions instead of raw terminal output. It is the default for supported agents, and one switch shows the raw terminal. See [native chat](https://runtallos.com/features/native-chat). ## What is an agent connection? An **agent connection** is Tallos's setup for one agent CLI: installing it with the vendor's verified command, signing in or storing an API key in secure storage injected only into that agent, and listing its models. See [agent connections](https://runtallos.com/features/agent-connections). ## What is diff review? **Diff review** is reading the exact changes an agent made, line by line, before they are merged. In Tallos you comment on any line of a workspace's diff and send all comments back to the agent as one batch. See [diff review](https://runtallos.com/features/diff-review). ## What is vibe coding? **Vibe coding** is building software by describing what you want to an AI agent and steering it with more prompts, judging the result by how the app behaves rather than by reading the code. The term was popularized in early 2025 by Andrej Karpathy. Read the [vibe coding guide](https://runtallos.com/vibe-coding). ## What is agentic coding? **Agentic coding** is software development in which AI agents act on the codebase — editing files, running commands and tests — instead of only suggesting code. Unlike vibe coding, it can include full review of every change. Read [what AI coding agents are](https://runtallos.com/what-are-ai-coding-agents). ## What is a context window? A **context window** is the amount of text a language model can consider at once, including your prompt, the files it has read and the conversation so far. When a session outgrows it, the agent can lose earlier details, which is why small tasks work better. See [the limits of AI coding agents](https://runtallos.com/what-are-ai-coding-agents). ## What is MCP (Model Context Protocol)? **MCP (Model Context Protocol)** is an open protocol, introduced by Anthropic, that lets AI applications and agents connect to external tools and data sources through a standard interface called an MCP server. Many agent CLIs can use MCP servers to reach databases, issue trackers and other systems. Read [what AI coding agents are](https://runtallos.com/what-are-ai-coding-agents). ## What is an agent skill? An **agent skill** is a packaged set of instructions, usually a `SKILL.md` file with optional scripts, that an agent loads when a task matches it. Tallos installs its own skills into your agents so they can drive Tallos with the `tallos` command. See [Tallos CLI and skills](https://runtallos.com/features/tallos-cli-and-skills). ## What is an automation? An **automation** is a prompt that Tallos runs with an agent on a schedule — hourly, daily, on weekdays, weekly or on a custom schedule — so recurring work like triage or maintenance starts without you. Saved squads can be scheduled the same way. See [automations](https://runtallos.com/features/automations). ## What is a remote workspace (SSH)? A **remote workspace** is a workspace whose git worktree and agents run on another machine you reach over SSH, while you keep the same Tallos interface on your computer. It is useful for heavy builds or a more powerful machine. See [remote workspaces](https://runtallos.com/features/remote-workspaces). ## What are hooks? **Hooks** are commands that run automatically at a specific moment. Agent CLIs like Claude Code run hooks at lifecycle events, Tallos uses status hooks to show whether an agent is working, waiting or done, and worktree setup hooks run commands such as `pnpm install` after a new workspace is created. See [parallel workspaces](https://runtallos.com/features/parallel-workspaces). ## What are agent permissions? **Agent permissions** are the rules that decide whether an agent asks before running commands or editing files. Tallos starts agents in manual mode by default; autonomous mode is opt-in, under Advanced on the welcome screen or in Settings → Agents → Agent Permissions. Read about [autonomy and permissions](https://runtallos.com/what-are-ai-coding-agents). ## Frequently asked questions ### What is the difference between an ADE and an IDE? An IDE is built for one person editing code in one working copy. An ADE is built for directing several AI agents at once, each in its own isolated workspace, with diff review before merge. ### What is the difference between an AI coding agent and an agent CLI? An AI coding agent is any program that completes coding tasks by acting on your project. An agent CLI is an agent that runs as a command-line program in your terminal, which is how most coding agents ship today. ### What is the difference between a git worktree and a git branch? A branch is a line of commits. A worktree is a separate folder on disk with a branch checked out, so several branches can be worked on at the same time without switching. ### Is a squad the same as agent orchestration? A squad is one form of orchestration: a leader agent coordinating member agents under limits you set, ending in your review. Orchestration is the broader idea of coordinating agents' work. ### Is vibe coding the same as agentic coding? No. Agentic coding is any development where AI agents act on the code. Vibe coding is the informal style where you do not read the code the agent writes. --- **Run your agents in parallel with Tallos** — Claude Code, Codex, Gemini and 30+ agents — each in its own workspace, with squads, chat, terminals and review in one app. Uses the subscriptions you already have. https://runtallos.com/signup (macOS 13+ · Windows 10+)