This article defines the terms that show up across Unstoppable Code and Unstoppable Cowork. It's a glossary, not a how-to — skim it once to get your bearings, then come back whenever a word is unfamiliar. The Unstoppable Code terms come first, followed by the Unstoppable Cowork terms.

The two products are siblings on one platform: Unstoppable Code is for developers, and Unstoppable Cowork is for non-technical work. Where a word means slightly different things in each, it's defined in both sections.

Unstoppable Code

Foundations

• Workspace — the project, or set of projects, you're working in. A workspace holds your chats, worktrees, and pipelines, and carries its own settings: default instructions, plugins, and environments.
• Worktree — an isolated branch and checkout of your repository where a single task runs. Because each worktree is separate, agents never overwrite each other's changes, and you can have several running at once.
• Chat — where you talk to an agent. In Code, a chat runs inside a worktree. (The app also has a separate Pipeline Assistant chat for building pipelines, and Cowork chats for non-technical work — see below.)
• Agent — an AI coding assistant — Claude Code, Codex, and more via Ollama, — that does the work in a chat or a pipeline.
• Plan — a step-by-step implementation plan an agent drafts before any code is written, shown with diagrams and diffs. You refine a plan with targeted edits (revising specific lines rather than regenerating the whole thing), and a plan tracked in the app can be shared for team comments before it runs.

Automation

• Pipeline — a configurable, multi-step workflow that chains agents, quality gates, and scripts, and can fire on CI, build, and source-control events. Pipelines run in one of three modes: Dispatch (you start it manually), Plan-ready (it runs from a plan you've already made), and Task-ready (you hand it a task and it writes its own plan first).
• Pipeline Assistant — a chat dedicated to building and debugging pipelines, separate from the chat you use for coding tasks.
• Hook — an automatic response that fires on a worktree or pull-request event — a failed pre-commit, a failed CI check, a merge conflict, a new pull-request comment — to keep work moving without you. Hooks are durable rules that run whenever their event occurs.
• Watcher — a monitor that pings a chat when something you're watching — a chat, a pipeline, or a hook event — next finishes, succeeds, or fails. A watcher only notifies you; it doesn't start the work, and it watches future events only.
• Automation — a scheduled or event-triggered activity that does work for you, such as running a pipeline or starting a chat on a schedule. (Contrast with a watcher, which only notifies you.)
• App Advisor — a way to drive the whole app in natural language. It can start chats, create worktrees, build pipelines, and set up watchers for you.

Configuration

• Skill / Plugin — a reusable unit of agent behavior you import (typically from a repository) and reuse. Manage your skills and plugins in one place and use them with either agent.
• Collection — a saved grouping of skills, plugins, and MCP servers you can apply together.
• Marketplace — a source for discovering and installing plugins and skills.
• MCP server — a connection that gives agents tools and access to outside services, such as Linear or Slack.
• Environment / Runtime — a set of environment variables and secrets, split by runtime (global, agents, pipelines). Runtimes can inherit values from one another.
• Lifecycle scripts — scripts the app runs automatically at set moments: setup (when a worktree is created), teardown (when it's deleted), and launch (how to run your app). You can add your own custom scripts too.

Working at scale & staying in control

• Status indicators — the markers that show what's waiting, what's running, and what's finished and needs your attention.
• Multi-repo workspace — several repositories grouped into one workspace, either permanently or joined on the fly, so agents can work across related repos at the same time.
• Cloud — where agents run. You can run them locally on your own machine or on the Unstoppable agent cloud.
• Team feed — a live, shared feed that keeps teammates in the loop and doubles as shared context for your agents.
• Audit trail — a complete record of every session, including live transcripts, costs, and usage across people, models, and pipelines.
• Model — the AI model and effort level powering a task. Models trade off speed, capability, and cost, so you choose the right one per task.

Unstoppable Cowork

• Project — a container for related work, with its own context, notes, and files. A project might be a campaign, a customer, or a report.
• Chat — a conversation with the assistant inside a project. Archive a chat when you're done.
• Assistant — the AI that works alongside you in Cowork. (It's the Cowork term for what Unstoppable Code calls an agent.)
• Instructions — standing guidance you set on a project so the assistant always follows it.
• Notes — your own notes, saved and available across the whole project.
• Prompt — a reusable snippet you save in a project to standardize a request you make often.
• Attachments — files you upload for the assistant to read and work with.
• Artifacts — documents and other outputs the assistant creates, saved to your workspace so you can find and reuse them.
• Context — the files and notes you pin so the assistant always references them.
• Plugins & connections — apps you link, such as Slack or Linear, so the assistant can take action, not just give advice.