Copilot CLI — A Field Guide for Claude Code Users

You already know how to vibe with an agent. This doc just maps what you know in Claude Code to what you'll find in GitHub Copilot CLI — so you're productive on day one.

Disclaimer: This is a personal/community guide, not official GitHub documentation. For the canonical source see docs.github.com/copilot. All info here is derived from publicly available docs and the CLI's own /help output.

---

TL;DR — The 30-second version

• Same kind of tool. Terminal-native AI agent that reads/edits files, runs shell commands, and works in a loop.
• Same brains available. /model lets you pick Claude Sonnet/Opus, GPT-5, and others.
• Different wrapper. Different slash commands, different defaults, deeper GitHub integration.
• Biggest wins over Claude Code: /delegate to a cloud agent, /fleet parallel subagents, /remote control from phone, /research, native GitHub MCP, scheduled prompts (/every).
• Biggest adjustments: approval prompts per command, different prompt feel, GitHub-centric workflow.

---

1. Launching & basics

# Install (macOS / Linux)
brew install copilot-cli
# or: curl -fsSL https://gh.io/copilot-install | bash
# or: npm install -g @github/copilot

# Launch in any folder
copilot

First run → /login to authenticate with your GitHub account (needs an active Copilot subscription).

---

2. The "muscle-memory" cheat sheet

You used to do this in Claude Code	In Copilot CLI you do this
claude to start	copilot to start
/login	/login ✅ same
/model to switch models	/model ✅ same
/clear to reset chat	/clear (abandons session) or /new (fresh convo, keep session)
/compact to shrink context	/compact ✅ same
/resume previous session	/resume ✅ same
/init to create project memory	/init ✅ same (creates Copilot instructions)
CLAUDE.md for project memory	CLAUDE.md, AGENTS.md, or .github/copilot-instructions.md — all are honored
/agents to manage subagents	/agent (browse) + /fleet (parallel) + /tasks
/mcp for MCP servers	/mcp ✅ same
/permissions	/allow-all, /add-dir, /list-dirs, /reset-allowed-tools
/cost / usage	/usage
/help	/help ✅ same
/exit	/exit ✅ same
/chrome (browser control)	❌ Not built-in — add via MCP (chrome-devtools-mcp or playwright-mcp)
/ide to connect editor	/ide ✅ same
/review for code review	/review ✅ same
/pr workflow	/pr ✅ same, more GitHub-native
@file to attach a file	@ to mention files ✅ same
#issue mentions	# to mention issues & PRs ✅ same
!command shell escape	! to execute shell ✅ same
shift+tab to switch modes	shift+tab ✅ same

---

3. Full slash command reference

🌐 Agent environment

Command	What it does
/init	Generate project instructions file for the repo
/agent	Browse and pick from available agents
/skills	Manage skills for extra capabilities
/mcp	Manage MCP servers (browsers, databases, Figma, custom tools)
/plugin	Manage plugins and plugin marketplaces

🤖 Models & subagents

Command	What it does
/model	Switch the underlying LLM (Claude Sonnet/Opus, GPT-5, etc.)
/delegate	Send this session to GitHub — Copilot opens a PR for you ⭐
/fleet	Run multiple Copilot subagents in parallel ⭐
/tasks	View and manage running tasks/subagents

💻 Code

Command	What it does
/ide	Connect to a VS Code workspace
/diff	Review the changes you've made so far
/pr	Create, view, or operate on PRs for current branch
/review	Run the code-review agent on your changes
/lsp	Manage language server config (TypeScript, Pyright, etc.)
/terminal-setup	Enable shift+enter for multiline input

🔐 Permissions

Command	What it does
/allow-all	YOLO mode — auto-approve everything
/add-dir	Add a folder to allowed paths
/list-dirs	Show allowed directories
/cwd	Show or change working directory
/reset-allowed-tools	Clear approved tool list

🗂️ Session

Command	What it does
/resume	Switch to a previous session
/rename	Name (or auto-name) the current session
/context	Show context-window token usage
/usage	Session metrics & stats
/session	View and manage sessions
/compact	Summarize history to free context
/share	Export session to markdown, HTML, or GitHub gist
/remote	Control this session from GitHub web/mobile ⭐
/copy	Copy last response to clipboard
/rewind (or /undo)	Undo last turn + revert file changes

🛟 Help & meta

Command	What it does
/help	Show interactive help
/changelog	See what's new (add summarize for an AI summary)
/feedback	Submit confidential feedback
/theme	Light / dark / custom
/statusline, /footer	Customize the status line
/update	Update the CLI
/version	Version info + update check
/experimental	Toggle experimental features (e.g. autopilot)
/clear	Abandon session, start fresh
/instructions	View / toggle custom instruction files
/streamer-mode	Hide model names & quota (for streaming/demos)

✨ Power-user commands

Command	What it does
/ask	One-off side question (doesn't pollute history)
/autopilot	Toggle autopilot — keep working until task is done ⭐
/chronicle	Browse session history + insights
/env	Show everything loaded: instructions, MCPs, skills, agents, plugins, LSPs
/every	Schedule a recurring prompt (e.g. /every 5m run tests) ⭐
/exit	Quit
/keep-alive	Prevent system sleep during long agent runs
/login / /logout	Auth
/new	Start a fresh conversation
/plan	Force a plan before writing code
/research	Deep multi-source research with citations ⭐
/restart	Restart CLI, keep session
/search	Search the conversation timeline
/sidekicks	View running sidekick agents
/undo	Same as /rewind
/user	Manage GitHub user list

⭐ = capabilities Claude Code doesn't have (or doesn't have as cleanly).

---

4. Keyboard shortcuts

Global

Keys	Action
/	Slash commands
@	Mention files
#	Mention issues / PRs
!	Run a shell command
shift+tab	Cycle modes (chat / agent / autopilot)
ctrl+s	Run command, keep input
ctrl+o / ctrl+e	Expand all timeline entries
ctrl+c	Cancel current action
ctrl+c ×2	Exit
esc	Cancel
ctrl+d	Shutdown
ctrl+l	Clear screen
ctrl+t	Toggle reasoning display
ctrl+x → b	Move current task to background
ctrl+x → o	Open most recent link

Input editing

Keys	Action
ctrl+a / ctrl+e	Start / end of line
ctrl+h	Delete previous char
ctrl+w	Delete previous word
ctrl+u / ctrl+k	Delete to start / end of line
meta+← / meta+→	Move by word
ctrl+g	Edit prompt in $EDITOR

---

5. Project memory & instructions

Copilot CLI respects all of these (in priority order):

CLAUDE.md                              ← yes, your existing Claude file works!
GEMINI.md
AGENTS.md                              ← in git root & cwd
.github/instructions/**/*.instructions.md
.github/copilot-instructions.md
$HOME/.copilot/copilot-instructions.md
COPILOT_CUSTOM_INSTRUCTIONS_DIRS       ← env var for extra dirs

Migration tip: your existing CLAUDE.md is honored as-is. No rewrite needed.

---

6. Five things that will surprise you (coming from Claude Code)

1. Approval prompts everywhere. Every shell command asks for OK. Use /allow-all for trusted dirs or /add-dir per folder. Speeds things up dramatically.
2. /delegate is magic. Hand off a task → it runs on GitHub's servers → opens a PR. No equivalent in Claude Code.
3. /fleet runs parallel agents. Multiple subagents on independent tasks at once. Big speedup for batch work.
4. /remote = phone control. Walk away, steer from GitHub mobile. Wild.
5. /every schedules prompts. "Every 10 minutes, run tests and tell me if anything broke." It's a cron for your agent.

---

7. What's missing vs Claude Code (be honest)

Claude Code has	Copilot CLI alternative
/chrome (Claude Computer Use browser)	Add chrome-devtools-mcp or @playwright/mcp via /mcp
Anthropic-tuned default tone	Switch /model to Claude Sonnet 4.5 / Opus
Artifacts-style live preview (claude.ai)	Use /ide to connect VS Code for visual feedback
/hooks for scripted automation	Use /every + MCP for similar workflows

---

8. Recommended first-day workflow

# 1. Install + launch
brew install copilot-cli
cd your-repo
copilot

# 2. Initial setup
/login                    # GitHub auth
/model                    # pick Claude Sonnet 4.5 if you want familiar feel
/init                     # creates instructions for this repo
/terminal-setup           # enable shift+enter multiline
/allow-all                # (optional) skip approval prompts in this folder

# 3. Try the unique stuff
/research                 # deep research
/delegate                 # send a task to cloud agent → PR
/fleet                    # parallel subagents
/every 10m run tests      # scheduled task

# 4. When stuck
/help                     # full command list
/env                      # see everything loaded
/feedback                 # tell the team what's broken

---

9. Quick reference card (print this)

START         copilot
HELP          /help
MODEL         /model          (Claude / GPT-5 / etc.)
FILES         @path/to/file
ISSUES/PRs    #123
SHELL         !ls -la
MODES         shift+tab
UNDO          /undo  or  /rewind
NEW CHAT      /new
SAVE TIME     /allow-all
BIG TASK      /delegate       (→ GitHub PR)
PARALLEL      /fleet
RESEARCH      /research
SCHEDULE      /every 5m <prompt>
PHONE         /remote
QUIT          /exit

---

Welcome to Copilot CLI. Same brain (if you want), bigger toolbox, deeper GitHub. 🚀