Skip to content

AI Agents

TermBeam can detect AI coding agents installed on your machine and launch them with a single tap — perfect for starting a Copilot, Claude, or Aider session from your phone.

AgentCommandWebsite
GitHub Copilotcopilotdocs.github.com/copilot
GitHub Copilot (gh)gh copilotgithub.com/github/gh-copilot
Claude Codeclaudedocs.anthropic.com
Aideraideraider.chat
Codex CLIcodexgithub.com/openai/codex
OpenCodeopencodeopencode.ai

TermBeam auto-detects which agents are available in your PATH on first request, with results cached for 60 seconds.

  1. Open the New Session modal (tap ”+ New Session”)
  2. Choose your working directory
  3. Tap an AI agent to launch — a new terminal session opens with the agent running

To resume a previous agent session, open the Command Palette (Ctrl+K or tap the ⌘ button) and select “Resume agent session”.

When you launch an agent, TermBeam creates a regular terminal session and runs the agent command inside it. This means:

  • Full agent features — You get the complete CLI experience (agent mode, tool calling, file editing, codebase awareness) — not a watered-down SDK version
  • All TermBeam features apply — TouchBar, themes, push notifications, split view, wake lock
  • Session management — The agent session appears in your session list like any other terminal
GET /api/agents

Returns detected AI agents:

{
"agents": [
{
"id": "copilot",
"name": "GitHub Copilot",
"cmd": "copilot",
"args": [],
"icon": "copilot",
"version": "1.2.3"
},
{
"id": "claude",
"name": "Claude Code",
"cmd": "claude",
"args": [],
"icon": "claude",
"version": "2.0.0"
}
]
}

Results are cached for 60 seconds. The endpoint requires authentication.

If no agents are detected, install one:

Terminal window
# Claude Code
npm install -g @anthropic-ai/claude-code
# GitHub Copilot (via gh CLI)
gh extension install github/gh-copilot
# Aider
pip install aider-chat
# Codex CLI
npm install -g @openai/codex

After installing, restart TermBeam or wait 60 seconds for the cache to refresh.

  • TouchBar keys work great with AI agents — Tab to accept suggestions, Esc to cancel, arrows to navigate
  • Push notifications will alert you when a long-running agent task finishes
  • Wake Lock keeps your screen on during agent work
  • Split View lets you run an AI agent alongside a regular terminal
  • Copy Overlay makes it easy to copy agent output on mobile

  • Usage Guide — sessions, panes, command palette, and the touch keyboard
  • Resume & List — reconnect to an active agent session from a desktop
  • Configuration — env vars and CLI flags