Termi Protocol

Product Introduction

1. Definition: Termi Protocol is a local-first, 3D visualization and orchestration platform for AI coding agents. It functions as a persistent, visual command center that runs on top of existing command-line interface (CLI) agents like Claude Code, Codex, and Gemini CLI.
2. Core Value Proposition: It exists to solve the opacity and context-loss problems inherent in terminal-based AI agent workflows. By providing a persistent 3D environment where every agent action (file reads, terminal commands, code edits) is visualized in real-time, it offers developers unprecedented visibility, control, and continuity over automated coding tasks.

Main Features

1. Live 3D Agent Visualization: Every action from a connected CLI agent is translated into a visual metaphor within a 3D room. Agents are represented as avatars that walk to file cabinets to read, sit at desks to write code, and execute commands in a visible terminal. This provides a non-scrolling, spatially intuitive log of the entire agent workflow.
2. Persistent Project Memory & State Management: Unlike a terminal session, Termi Protocol automatically saves the complete state of a project. This includes agent goals, task boards (kanban), code change checkpoints, and learned project context ("memory"). Users can close the application and resume work exactly where they left off, eliminating the need to re-explain context.
3. Multi-Agent Orchestration & Handoff: The platform includes a "boss agent" capable of splitting a single goal across a team of specialized AI agents. Crucially, it features automated cross-model handoff: if one agent hits a usage, rate, or context limit, Termi saves its progress (goal, changed files, errors, todos) and seamlessly hands the work to an agent using a different AI model (e.g., from Claude Code to Gemini CLI).
4. Integrated Command Center Per Agent: Each agent's desk is backed by a comprehensive control panel featuring a genuine xterm/PTY shell (not a log replay), a live cost tracker (tokens & USD), the interactive task board, snapshot checkpoints for rewinding changes, and the agent's accumulated project memory. File locking prevents collisions between multiple agents.
5. Gamified Productivity Layer (Moscow Protocol): The platform incorporates a game layer where real coding workflow steps earn experience points (XP). Users progress through leagues (Rookie to Legend) and can appear on a global leaderboard, adding a meta-layer of engagement and performance tracking to AI-assisted development.

Problems Solved

1. Pain Point: Terminal Opacity and Loss of Control. Standard CLI agents operate as a "black box" with only a scrolling text log for feedback, making it difficult to understand their reasoning, approve changes confidently, or intervene before costly errors.
2. Pain Point: Session Fragility and Context Loss. Closing a terminal window destroys the agent's thread state. Developers must manually save prompts and re-explain complex project context for every new session, wasting time and breaking flow.
3. Target Audience: Software Engineers and Developers who regularly use AI coding assistants (Claude Code, Cursor, Aider, etc.) for complex tasks like feature implementation, debugging, or refactoring. It is particularly valuable for tech leads and solo founders managing multi-agent workflows or those frustrated by the limitations of chat-based AI coding.
4. Use Cases: Orchestrating a multi-agent refactoring project where one agent handles backend logic, another the frontend, with a boss agent coordinating. Resuming a complex, multi-day bug fix with full context intact. Monitoring and approving an agent's proposed changes to a critical codebase with visual confidence. Handling AI provider outages or limits by automatically failing over to a different model without losing progress.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • Unmatched Visibility: Transforms abstract terminal logs into intuitive, watchable 3D actions.
     • True State Persistence: Solves the fundamental "amnesia" problem of CLI agents between sessions.
     • Practical Multi-Agent Workflow: Cross-model handoff and file locking are unique, production-ready features for reliable automation.
     • Local-First & Private: Uses user's own API keys and runs agents locally; code never leaves the machine.
     • Clear, Transparent Pricing: One-time lifetime license with no subscription or usage caps.
   • Cons:
     • Learning Curve: Introduces a new paradigm (3D room) which may be initially unfamiliar compared to a pure terminal.
     • System Resource Overhead: Running a 3D visualization layer will consume more RAM/GPU than a headless CLI, potentially impacting performance on lower-end machines.
     • Dependent on External Agents: Core functionality is tied to the stability and capabilities of third-party CLI agents (Claude Code, etc.).
     • Early-Stage Roadmap: Many promised features (Istanbul through Giza protocols) are on a future roadmap and not yet available.

2. Key Alternatives & Differentiation:

   • Raw CLI Agents (Claude Code, Aider): Termi Protocol is not a competitor but a visualization and management layer for these tools. Differentiation: CLI agents offer zero visualization, no persistent state between runs, and no multi-agent orchestration. Termi adds these on top.
   • AI-Powered IDEs (Cursor, Windsurf): These integrate AI directly into the code editor. Differentiation: Termi is agent- and model-agnostic, orchestrating external CLI agents. It focuses on workflow visualization, state persistence, and multi-model orchestration rather than deep editor integration.
   • Agent Orchestration Platforms (LangChain, AutoGen): These are developer frameworks for building custom agent workflows. Differentiation: Termi is an end-user desktop application with a pre-built, opinionated UI and 3D visualization. It trades low-level framework flexibility for a turnkey, visually managed experience.

Frequently Asked Questions (FAQ)

1. How does Termi Protocol handle my API keys and code privacy? Termi Protocol operates on a local-first architecture. Your AI API keys are used by your own CLI agents (e.g., Claude Code) running on your machine. Termi visualizes the output but does not transmit your code or keys to its own servers, ensuring full privacy.
2. Can I use Termi Protocol with any AI model or coding agent? Yes, Termi Protocol is designed to be CLI-agent agnostic. It officially supports popular agents like Claude Code, Codex, and Gemini CLI, but its architecture allows it to work with any command-line AI coding tool that can be run in a terminal, including local models via Ollama.
3. What happens if an AI agent gets stuck or goes into a loop? The live 3D visualization allows you to see the agent's activity in real-time. You can manually pause or stop the agent at any step from the Command Center. Furthermore, you can rewind to any previous checkpoint to revert bad edits or commands instantly.
4. Is the "Lifetime Pass" a subscription, and what does it include? No, it is not a subscription. The one-time $6.99 Lifetime Pass is a permanent license for the Termi Protocol application. It includes unlimited use of all current features (Milano protocol) and all future protocols and chapters (Istanbul through Giza) as they are released, with no recurring fees.
5. What are the system requirements for running Termi Protocol? Termi Protocol is a desktop application available for macOS, Windows, and Linux. As it renders a real-time 3D environment, it requires a system capable of running modern OpenGL or Vulkan graphics. Specific minimum RAM and GPU requirements are not listed but are higher than for a standard terminal application.