Agent Deck is an open-source terminal session manager for AI coding agents – a single command center that lets you run, watch and switch between many Claude Code, Gemini CLI, OpenCode, Codex and Aider sessions at once.
If you routinely keep a dozen agent sessions alive across different projects, Agent Deck replaces the tangle of terminal tabs with one TUI that shows every session’s state – running, waiting for input, or done – and switches between them with a keystroke. Sessions can be grouped, searched and forked (branching a conversation to try a different approach), and each can live in its own git worktree so parallel agents never collide. Its Conductor mode goes further: a supervising session watches the fleet, answers routine questions, and escalates the interesting ones to your phone through Telegram or Slack. There is also an MCP manager, a skills manager, a cost-tracking dashboard, Docker-sandboxed sessions and a web mode with a fleet board. Written in Go and released under the MIT license, Agent Deck has about 1,130 stars on GitHub and ships new releases at a rapid pace, with its latest push in October 2026.
Agent Deck is free and open source under the MIT license. You only need the underlying agent tools and model accounts you already use.
Agent Deck is for developers who run many AI coding-agent sessions in parallel and want mission control – visibility, quick switching and light orchestration – instead of a screen full of terminal windows.