Warp Agent CLI Cheatsheet - In-Terminal AI Coding Agent

All Warp Agent CLI terminal AI coding-agent commands in one place — from install to core agent commands, MCP integration to cloud agents — organized by scenario. Copy and go.

AI CLI·17 commands·Last updated 2026-08-26

Install & Launch 4

brew install --cask warp
Install Warp on macOS via Homebrew
sudo dpkg -i warp-terminal_*.deb
Install on Debian/Ubuntu with .deb
warp
Launch the Warp terminal (AI features need sign-in)
Ctrl+I
Open Agent Mode in any Warp session

Core Agent Commands 7

/agent [prompt]
Start a new conversation
/model
Switch the base agent model
/plan [task]
Ask the agent to research and plan a task
/orchestrate [task]
Split a task into subtasks and run them with parallel agents
/compact [instructions]
Summarize history to free context
/clear [prompt]
Clear the transcript and start fresh
/conversations
Open conversation history

Integrations & Cloud Agents 6

/mcp
View and manage MCP servers
/skills
Browse in-scope skills and insert one
/export-to-file [filename]
Export the current conversation as a Markdown file
oz login
Sign in to your Warp account (cloud agent CLI)
oz agent run --prompt "Add a changelog entry"
Run a cloud-agent task locally
oz agent run-cloud
Dispatch a headless run to Warp cloud

Tips

  • Warp Agent is agent-first: Enter sends input to the agent by default; press `!` to enter shell mode and run commands directly, switching back and forth.
  • `/orchestrate` splits a big task into subtasks run by parallel agents — good for several independent jobs at once.
  • For long tasks or CI, `oz agent run-cloud` dispatches a headless run to Warp cloud so your local terminal need not stay attached.

FAQ

Is the Warp Agent CLI a standalone program or part of the terminal?

Agent capabilities are built into the Warp terminal (Rust). Launch `warp`, then use Agent Mode (Ctrl+I) or slash commands like `/agent`. There is also a standalone `oz` CLI to dispatch tasks to Warp cloud.

How is Warp Agent Mode different from a normal chatbot?

It works through the CLI abstraction: it reads/writes files, runs shell commands, and calls any CLI/API with --help, asking your approval at each step and self-correcting — not just codegen or chat.

Official References

Each command links to its official documentation below, so you can verify the latest usage and read deeper.

Maintained by LaoHand

Publicly updated on Aug 26, 2026, continuously proofread against official docs.

Contact Us

Wrong command or description? Send us corrections, business inquiries or product feedback by email.

Contact Us

Edit this page on GitHub

Edit the source file directly and open a PR to improve this cheatsheet.