
Termi Protocol
Watch AI agents code in 3D and hand off the work
Sign in to rate this tool.
Overview

Termi Protocol lets you watch AI coding agents perform every file read, write and terminal command in a three‑dimensional workspace. The interface shows each step as a physical action in a virtual room, so you can see exactly what the agent does without scrolling through logs.
The tool works with any CLI agent you already use, such as Claude Code, Codex, Gemini CLI, Copilot, Grok CLI, Aider, Antigravity, opencode, Ollama, Cursor, Groq, or any other agent you run on your local machine. You bring the agent into Termi, and it assigns the agent a desk and a command center in the 3D room. The real shell (xterm + PTY) renders live terminal output. The agent's actions, moving to a cabinet to read a file, sitting at a desk to write code, or issuing a command, are mapped to 3D motions that you can watch in real time.
With Termi you can pause, approve, or reorder steps. A checkpoint system records every change, letting you rewind to a chosen point in seconds. Each agent has a view‑passport showing its name, model, goal, live token use and current cost. The command center displays live cost, tasks, checkpoints and memory for that agent. You can keep file locks so two agents never edit the same file simultaneously. The app saves task boards, checkpoints and project memory between sessions, so closing the room does not erase context.
Termi provides a full command center for every agent, a kanban board that the agent drives, a real terminal for real command execution, and live cost tracking. The workspace can be customized with desks, focus tools, and even a pet that grows with your coding days. A leaderboard tracks XP earned by completing real workflow steps, adding a gamified layer to the process.
Key features
- 3D visualisation of terminal actions for read, write and command steps
- Real shell integration (xterm + PTY) for live execution
- Pause, approve, and rewind controls with checkpoint snapshots
- Pass‑port hover for agent name, model, goal, tokens and cost
- File lock system to prevent concurrent edits
- Persistent task boards and project memory between sessions
Details
- Pricing
- Free
- Last checked
- 19h ago
- Indexing
- noindex, follow
Suggest an edit
Sign in to suggest an edit.
Pros and cons
Pros
- Full visibility of AI agent actions without scrolling logs
- Real-time cost monitoring per agent and session
- Persistence of context and checkpoints across sessions
- Compatibility with any local CLI agent
- Customisable 3D workspace and gamified XP system
Cons
- No published pricing or subscription details aside from one‑time Lifetime Pass
- Requires users to bring their own AI models or agents
- Limited information on integration with external project management tools

