Cursor CLI: Headless, Terminal-Native Coding (2026)

The CLI is Cursor's coding agent without the editor window. It runs in a terminal, so you can script it or run it headlessly in CI. Interactive editing and visual diff review still belong in the editor. Reach for the CLI for automation and for long runs that nobody watches.

Type agent in any shell to open the Cursor agent. The slash-command menu is your control surface from there: it lists the models you can switch to, your skills, the built-in commands and any MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. servers you have set up.

Prefix a message with & to hand the conversation to the cloud. The CLI starts a cloud agent that works on its own branch, asynchronously, so you can keep typing locally while it runs elsewhere. To run a shell command without leaving the session, enter Shell Mode with /shell.

/open opens the repository's Git root in the Cursor editor, where you can review the agent's changes visually, and /cursor is an alias for it. Cursor's docs describe it as opening the repository and don't say the conversation moves with it.

Shift+Tab rotates between the Agent, Plan and Ask modes without leaving the prompt, so you can drop into Plan before a big change or switch to Ask when you only want answers. Debug modeA mode that diagnoses a failure: it reproduces the issue, adds instrumentation and watches the logs, rather than reviewing a pull request. Press Enter for the full definition. and Run Everything, which runs every tool call without asking, sit outside that cycle, and each has its own slash command.

The slash menu also handles MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. servers, including the MCP login flow.

Yes, and the CLI makes it easy. Open a terminal tab per model, start an agent in each and point them all at the same directory. One tab runs Cursor's in-house ComposerCursor's own fast coding model, tuned for the editor and priced well below frontier models; the recommended day-to-day model for executing a plan. Press Enter for the full definition. model, another a flagship model such as GPT-5.x, and a third an Opus model. Give every tab the same brief, and you have parallel agents producing competing implementations of the same thing.

Switch a tab's model from the slash menu, where you can also set its reasoning level. It's a fast way to compare models on real work instead of arguing about which is best in the abstract. Name each tab's output so the runs don't overwrite each other, or give each run its own worktree. Let all three finish, then read the diffs side by side and keep the one you trust.

agent -p "..." runs the agent headlessly: a prompt goes in and a result comes out, with no interactive UI. Use that form in CI jobs, git hooks and shell scripts, where nobody sits at a keyboard to approve steps. Pair it with tight permissions and a review gate, as you would any unattended automation. Cursor's docs disagree on whether it edits files by default: the headless guide says changes are only proposed unless you add --force, and the usage page says the agent has full write access in non-interactive mode. Test it on a throwaway branch before a pipeline depends on either behavior.

The Cursor agent isn't tied to one editor. Through ACPAgent Client Protocol. An open standard that lets any IDE host any coding agent; Cursor uses it to run inside JetBrains/IntelliJ and Android Studio. Press Enter for the full definition. it runs inside JetBrains IDEs such as IntelliJ and PyCharm, with the CLI's core Agent, Plan and Ask modes and the same models. A team that works in JetBrains IDEs can use the Cursor agent without switching editors.

ACPAgent Client Protocol. An open standard that lets any IDE host any coding agent; Cursor uses it to run inside JetBrains/IntelliJ and Android Studio. Press Enter for the full definition. is the Agent Client ProtocolAn open standard that lets any IDE host any coding agent; Cursor uses it to run inside JetBrains/IntelliJ and Android Studio via the Cursor CLI bridge. Press Enter for the full definition., an open standard for coding agents: any IDE that implements it can host any agent that implements it. On the Cursor side, ACP runs through the CLI. The same binary that runs agent in your terminal also runs the ACP bridge, so the JetBrains and Android Studio integrations are the CLI with a different front end.

MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. config, slash commandsSaved multi-step prompts you trigger from the / menu, such as run tests, fix CI or open a pull request. Press Enter for the full definition., rules, skills, sub-agents and model selection mostly work the same way, so learning the CLI teaches you most of the ACPAgent Client Protocol. An open standard that lets any IDE host any coding agent; Cursor uses it to run inside JetBrains/IntelliJ and Android Studio. Press Enter for the full definition. surface. Cursor's ACP docs show where it differs: sessions support the three core modes, which leaves out Debug modeA mode that diagnoses a failure: it reproduces the issue, adds instrumentation and watches the logs, rather than reviewing a pull request. Press Enter for the full definition., and team-level MCP servers configured in the Cursor dashboard aren't supported in ACP mode.

Install the JetBrains AI Assistant pluginA Cursor marketplace package that bundles MCP servers and skills (sometimes sub-agents and hooks); one click installs all of it into your Cursor instance. Press Enter for the full definition., then add Cursor from the ACPAgent Client Protocol. An open standard that lets any IDE host any coding agent; Cursor uses it to run inside JetBrains/IntelliJ and Android Studio. Press Enter for the full definition. registry in the IDE's AI chat. Cursor's JetBrains guide gives the steps: open the agent provider list, choose Add Agent from Registry, search for Cursor, install it and select it as your agent provider. The guide lists a paid Cursor plan as a prerequisite and doesn't list a separate CLI install. It names JetBrains IDEs such as IntelliJ and PyCharm, and it doesn't mention Android Studio by name.

When you wire Cursor into CI or a custom tool, treat the CLI reference pages as the contract. The overview explains when the CLI is useful. The reference pages pin down what a script depends on: install commands, print mode, output formats, authentication, configuration and permissions.

Run /context to see what is using the context window, broken down by category. Check it as a session grows, because quality starts to slip well before the window is technically full.

Good fits are failure triage in CI and longer refactors that you start from a script and walk away from. Because the CLI is terminal-native, it slots into existing automation, but keep a human review gate. The CLI produces the changes that reach review, and merging them stays a deliberate, human step.

Scope what the agent may do before you let it run unattended: confine it to a branch, set permissions and shell access on purpose, and require review of its output. Give it more autonomy only once your own codebase gives you evidence that it's safe. Check the official CLI docs for the current commands, flags and permission model before you wire it into a pipeline.