TUIOS Terminal Workspace
Manage panes, sessions, and coding agents in a recoverable terminal workspace.
- Source repo
- Gaurav-Gosain/tuios
- Stars
- ★ 5k
- Last updated
- today
- License
- MIT
- Primary language
- Go
- FA score
- 59/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodex · Claude Code (Partial support)
- Cost
- Free, no paid service needed
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- Developers running builds, tests, and shells side by side can tile panes and separate tasks into workspaces.
- Not a fit if
- Users who work only in graphical interfaces and do not need terminal workflows
- Users who need full image and animation support in a basic terminal
- Source review
- 59/100 · Major gaps
What does this agent do, and when should you use it?
TUIOS is a Go terminal multiplexer and window manager that runs inside an existing terminal, with Vim-style modal controls, multiple panes, workspaces, and BSP tiling. A daemon keeps sessions alive and supports detaching, reattaching, and restoring session layouts after daemon or machine restarts. Coding agents can report their state through integrations, while an Inbox collects approvals, questions, errors, and completion notices across sessions and machines. Other programs can report status through OSC 7501 or `tuios status`, and agents can message or question one another through `tuios` commands. The project also documents SSH remote sessions, a separate web-terminal binary, and an MCP server mode; graphics features depend on terminal support.
TUIOS starts shells or other commands in terminal panes that users can create, resize, drag, and tile, then organize with workspaces, layout modes, and session commands. Its daemon preserves session structure and working directories and supports detach/reattach and access across machines. tuios integration install wires 19 harnesses for state reporting, and TUIOS also detects 24 agent CLIs by process and screen; state appears in pane titles and the sidebar. Other programs can report state with tuios status or OSC 7501. The Inbox gathers approvals, questions, mail, errors, and completed turns, and users can respond to prompts there. tuios fan starts several agents in separate Git worktrees, tuios send-agent-message and tuios ask-agent support agent communication, and tuios mcp exposes a read-only MCP interface limited to the agent's own session by default.
- Developers running builds, tests, and shells side by side can tile panes and separate tasks into workspaces.
- Individual developers who often close terminals or reboot can use daemon sessions to retain pane structure and working directories.
- Developers running Claude Code, Codex, Gemini CLI, or opencode can surface agent state and pending approvals in the Inbox.
- Engineers working across machines can add SSH hosts, attach remote sessions, or run panes on remote hosts.
- Maintainers who repeat terminal workflows can record, replay, or execute workflows against a running session with tape scripts.
How do you install or deploy this agent?
Install with Homebrew; other documented options include Arch AUR, Nix, go install, Docker, and release binaries. Building from source requires Go 1.26.6 or newer. The runtime requirement is a terminal with true color support; Kitty graphics and Sixel are recommended. SSH is needed for SSH host features. The README does not state a price; its listed install and run methods do not require model API keys.
brew install tuiosGo installation:
go install github.com/Gaurav-Gosain/tuios/cmd/tuios@latestBuild from source:
git clone https://github.com/gaurav-gosain/tuios.git
cd tuios
go build -o tuios ./cmd/tuios
./tuiosHow do you use this agent?
Run tuios to launch; by default it attaches to a daemon-backed session. Create and reattach to a persistent session, or list sessions with tuios ls. The default prefix key is Ctrl+B, and keybindings can be configured.
tuiosPersistent session example:
tuios new mysession
tuios attach mysession
tuios lsAfter installing agent-state integrations, use the Inbox to review pending items; tuios mcp starts the MCP server. Configure remote hosts with tuios hosts add, then connect to a remote session with tuios attach --host. Exact setup options vary by target agent and host policy; the README does not give one universal credential checklist.
What are this agent's strengths and limitations?
- Daemon sessions support detach and reattach, and can restore session structure and working directories after restarts.
- The Inbox consolidates agent approvals, questions, messages, errors, and completion states across sessions and machines.
- It can start multiple agents in separate Git worktrees and includes commands for agent-to-agent questions and messages.
- Event-driven rendering is described as using zero CPU while idle.
- Users need to learn Vim-style modal controls and a configurable prefix key, which defaults to Ctrl+B.
- Kitty graphics and Sixel depend on terminal support; Sixel is experimental and lacks pixel-level clipping.
- Multi-machine sessions require SSH host and policy configuration.
- Building from source requires Go 1.26.6 or newer.
How does this agent compare with similar options?
The README presents TUIOS as a tmux alternative and cites tmux, vim, and i3 as design influences. Beyond terminal-pane management, TUIOS adds agent state, a cross-session Inbox, agent messaging, and cross-machine sessions; whether it fits as a replacement depends on whether you need those integrations and their configuration.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| TUIOS Terminal Workspace This agent | 59 · Major gaps | CLIFree | ★ 5k | today | Go | — |
| tty7 Terminal Workbench | 67 · Some gaps | Desktop appFree | ★ 1.2k | 2d ago | Rust | — |
| cmux | 47 · Major gaps | Desktop appFreemium | ★ 28k | today | Swift | Codex · Claude Code |
| Runner Multi-Agent Terminal | 82 · Some gaps | Desktop appFree | ★ 272 | today | Rust | Codex · Claude Code |
How does FollowAgents rate this agent?
Why each dimension lost points
The README describes pane grants, an MCP surface that is read-only and scoped to the agent's own session by default, and an Inbox approval flow, giving partial evidence of permission and confirmation controls. Hooks can run shell commands, tools can act across machines, and the install script fetches and installs software; the supplied material does not show the detailed authorization path, sensitive-data protections, recovery from external effects, or dependency audit results, so those scores remain limited. The security contact and acknowledgments provide some attribution, but publisher identity is unverified and earns no brand-based credit.
The README's descriptions of daemon sessions, restart recovery, and updates are broadly consistent. It also gives release checksums, a Go version requirement, and several installation routes. The supplied material does not show runtime failure-message design or independent verification of these claims, so failure messages score 1 and the other criteria are not full marks.
The documentation names concrete scenarios including terminal multiplexing, coding agents, remote machines, and automation. It also states terminal graphics requirements, grant controls, the read-only MCP scope, and limits for some experimental features. Audience and boundaries are reasonably clear, though evidence about the full constraints across environments is limited.
The README has clear navigation, a quick start, command examples, and links to focused guides. Installation routes, the MIT license text, vulnerability reporting, prerelease support policy, and v0.8.5 notes are included. The supplied material does not establish a FAQ, a systematic changelog, or stronger maintenance commitments; SECURITY.md says maintenance is in personal free time with no response-time guarantee, so maintenance responsibility scores 2.
The stated combination of terminal window management with coding-agent status, Inbox, approvals, messaging, and remote sessions has clear potential value, and the README includes usable key and CLI examples. Its cost-benefit depends on Go 1.26.6, terminal capabilities, and daemon or remote configuration; the supplied material includes no observed usage evaluation, so this is not a full score.
The README links features to focused guides, CLI and configuration references, and architecture material, and labels some limitations as experimental. However, the supplied evidence is mostly README claims, without corresponding code or test excerpts to corroborate core features; no runtime results were executed in this static review. Traceability and fact/inference separation therefore receive moderate scores, while cross-source corroboration is low.
- This is a low-confidence static review: the supplied material contains no implementation or test files to verify README feature claims, and no code was executed.
- Before using shell hooks, cross-machine actions, or the install script, inspect their commands, permissions, and data flows. The security policy supports only the latest default-branch commit and gives no response-time guarantee.
- Terminal graphics support and the Go version requirement affect usability; Sixel and Kitty text sizing are marked experimental or as having known issues.
FAQ
Does TUIOS require a paid model or API key?
Do sessions survive closing a terminal or restarting?
Which coding agents can report state?
tuios integration install configures 19 harnesses, and TUIOS also detects 24 agent CLIs by process and screen. See the agent guide for specific names and integration methods.