This is an early release preview. You may encounter bugs.
Agent Manager logo
Unclaimed

Tool coding interface

Agent Manager

Terminal UI to manage AI coding-agent sessions (Claude Code, OpenCode, Codex, Grok, Gemini CLI) in tmux: live status, group tree, quick prom

A 81/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
Discussion

01 / Summary

About Agent Manager.

Agent Manager is a terminal UI that runs many AI coding-agent CLIs side by side, each in its own persistent tmux session, and shows them in one list with live status. It is a thin layer over the tools you already have installed: each session launches your own CLI as-is, so your login, subscription, configuration files, MCP servers, and the tool's own features carry over unchanged.

Sessions appear in a project tree that you can fold and reorder, with a preview of each agent's latest answer. You can send a prompt into any session without attaching to it, spawn a new agent in a group, revive a dead session on its own conversation, fork a conversation into a separate named session, or restart one on an empty context. Each session can start on its own model, reasoning effort, and profile, read from the CLI at runtime, and can optionally run in its own git worktree. Sessions live on a private tmux server, so they survive the manager quitting and never mix with your own tmux sessions.

A diff review panel shows a session's changes as full-file diffs, where you comment on lines and send numbered review rounds back to the agent. Shell terminals for builds and Git sit in the same list under an agent or group. Every MCP-capable session gets tools to spawn another agent, message it, and wait for it to finish. Agent Manager runs on macOS and Linux, and on Windows inside WSL2. Cost tracking is not yet available.

  • Supported CLIs: Claude Code, Codex, OpenCode, Grok Build, Gemini CLI, Antigravity CLI, Pi, Command Code, Hermes Agent, Muse Code, and Oh My Pi
  • Live status: one list shows which agents are working, done, or stuck
  • Quick prompts: answer a session or spawn a new agent without attaching
  • Revive, fork, and restart: bring back a killed session on its conversation, branch it into a named fork, or start fresh
  • Per-session model and effort: pick model, reasoning effort, and profile from the options each CLI reports
  • Git worktrees: optionally give each session its own worktree and branch
  • Diff review: full-file diffs with line comments sent back to the agent in numbered rounds
  • Agent-to-agent tools: MCP-capable sessions can spawn, message, and wait on other agents
  • Terminals alongside agents: shells nested under an agent or group for builds and Git
  • Mouse and keyboard control: drag to reorder or regroup rows, plus remappable keybindings
  • Settings: themes that follow the OS light or dark mode, list density, desktop notifications, and editor choice

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

04 / Build

Build with Agent Manager.

Browse the catalog for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get Agent Manager →

Browse the catalog