# mAgent

> mAgent is a free, open-source CLI that opens each of your projects in its own terminal window, launches your AI coding agent inside it, and auto-tiles every window into a grid across all of your monitors. One command — `magent --go` — takes you from an empty desktop to a fully tiled, resumed AI agent workspace.

Key facts:

- Product name: mAgent. Package: magent-multi-ai-agents-manager; CLI command: magent (`pip install magent-multi-ai-agents-manager`, run `magent` or `magent --go`).
- Agents it launches and tiles: Claude Code, OpenAI Codex CLI, Cursor CLI (cursor-agent), Antigravity, plus editors like VS Code and Cursor — anything on PATH.
- Live agent state and automatic conversation resume are wired for Claude Code and Codex only; other tools launch fresh.
- Sessions auto-resume: `claude --continue` and `codex resume` per window, restoring the same grid every time.
- `magent watch` shows a live table of every agent session — working, needs-input, done, error — sorted most-urgent first; `magent status --json` for scripts and CI. Wire the state hooks once with `magent hooks install` (Claude Code settings + the one-line Codex recipe); the press-a-row-number-to-focus action is Windows-only.
- Remote: on Windows, sessions live in psmux and survive disconnects; attach over SSH/Tailscale. On every platform you can pair Happy for end-to-end-encrypted phone/browser control. The upload server binds loopback + your Tailscale address by default, never the LAN wildcard unless you pass `--host`.
- Platforms: Windows, macOS, Linux · Python 3.10+ (some power features are Windows-only).
- License: AGPL-3.0, by DevinoSolutions.
- No credentials, API key, MCP server, or SDK required — mAgent is a local CLI only.

## Agent setup

- An AI coding agent can set mAgent up for itself: fetch and execute [/agent-setup/prompt.md](https://magent.now/agent-setup/prompt.md) — plain markdown, no auth, `Content-Type: text/markdown`.
- [/agent-setup/](https://magent.now/agent-setup/): human-readable per-agent setup guides (Claude Code, Codex, Cursor, OpenCode) the prompt is generated from.

## Pages

- [Home](https://magent.now/): overview of tiling, resume, watch table, remote control, and platform capability matrix
- [Install guide](https://magent.now/install): step-by-step install — requirements, pip/uv/source install, first-run config, agent state hooks, optional extras, `magent doctor`
- [FAQ](https://magent.now/faq): frequently asked questions
- [Blog](https://magent.now/blog): guides and workflows for running AI coding agents in parallel

## Blog posts

- [How to run multiple Claude Code sessions in parallel](https://magent.now/blog/run-multiple-claude-code-sessions)
- [mAgent vs tmux: which fits an AI agent workflow?](https://magent.now/blog/magent-vs-tmux-for-ai-agents)
- [Auto-resume Claude Code and Codex sessions after a restart](https://magent.now/blog/auto-resume-claude-code-codex-sessions)
- [Control your AI coding agents from your phone](https://magent.now/blog/control-ai-coding-agents-from-your-phone)
- [Live status for every AI agent session with lifecycle hooks](https://magent.now/blog/ai-agent-status-lifecycle-hooks)
- [Managing a fleet of AI coding agents across projects](https://magent.now/blog/parallel-ai-coding-agents-workflow)
- [Tile terminal windows across monitors, automatically](https://magent.now/blog/tile-terminal-windows-across-monitors)
- [One command to set up your entire dev workspace](https://magent.now/blog/multi-project-development-setup)

## Source

- [GitHub repository](https://github.com/DevinoSolutions/magent-multi-ai-agents-manager)
- [PyPI package](https://pypi.org/project/magent-multi-ai-agents-manager)

## Full text

- [llms-full.txt](https://magent.now/llms-full.txt): this index plus the complete text of every article above, inlined in one file.
- Every article is also served as clean markdown at its own URL with `.md` appended, e.g. https://magent.now/blog/magent-vs-tmux-for-ai-agents.md
- [RSS feed](https://magent.now/feed.xml) for new posts.