GitHub Copilot CLI brings GitHub’s coding agent to the terminal: it plans and executes multi-step build, edit, and debug tasks, previewing every action before it runs. It replaced the older gh-copilot extension, which only explained and suggested commands.
models and pricing
Copilot CLI switches with /model across Claude, GPT, Gemini, and xAI
models. Every Copilot plan includes the CLI, from the limited Free tier
through Pro at $10/mo, Pro+ at $39/mo, and Max at $100/mo to Business
and Enterprise seats, with usage counted in AI credits. It also connects
to your own model provider with your own key, and it runs local models
through Ollama.
what it’s good at
The draw is native GitHub integration: repos, issues, and pull requests
are addressable in natural language, and GitHub’s own MCP server ships in
the box alongside custom MCP support. It reads AGENTS.md, CLAUDE.md, and
copilot-instructions.md, supports custom agents with automatic task
delegation, and can run a session in a cloud sandbox with --cloud.
GitHub Copilot CLI in crystl
GitHub Copilot CLI runs in a crystl shard with its own model selection, activity visibility, and real approval cards for the tools it gates itself. Crystl treats its compatibility as partial because an allow-all session drops every card, and crystl surfaces that state instead of pretending to gate it.
The profile describes what crystl has tested; it never supplies a command, hook script, or launch setting. Those remain local to the Crystl installation.