📦

Overview

IATV (Independent Agent Terminal Viewer) is a PyQt6 control center for spawning, monitoring, and coordinating multiple Claude Code agents in independent terminal windows, with automatic agent creation on incoming messages and AfterLife reincarnation for uninterrupted long-running sessions.

The problem: Running multiple AI agents means juggling terminal windows, manually starting each one, losing track of which agent is working on what, and having no way for agents to trigger each other. When one agent needs help from another specialist, a human has to manually start that agent and relay the request.

The solution: IATV provides a visual command center for managing a fleet of Claude Code agents. Each agent runs in its own terminal window with a dedicated purpose (e.g., EE for infrastructure, MM for media, MacR for the desktop app). The control panel shows all running agents, their status, and lets you launch, monitor, minimize, or close any agent with a click.

Auto-Creation: The Killer Feature

When an agent sends a gmsg message to another agent that isn’t running, IATV automatically spawns that agent’s terminal window, delivers the message as a startup prompt, and the new agent begins working immediately.

This creates an organic, event-driven multi-agent system where agents wake each other up as needed — no human orchestration required.

Combined with AfterLife Reincarnation, agents that approach their context limit automatically save state and respawn. IATV manages this lifecycle transparently: the terminal window stays open, the new agent instance picks up the task, and the user sees uninterrupted progress.

Key Features

  • Agent Launcher - Spawn any configured agent (EE, MM, MacR, PubliKit, etc.) with one click
  • Terminal Monitoring - Live view of all agent terminal windows with status indicators
  • Auto-Spawn on Message - Agents that receive a gmsg message are automatically created if not running
  • AfterLife Integration - Agents reincarnate transparently when context fills up
  • gmsg Messaging Hub - Inter-agent communication routed through IATV’s message broker
  • Window Management - Zoom, minimize, close, arrange agent terminal windows
  • Agent-Specific Icons - Visual identification of each agent’s role in the fleet
  • Dark Theme UI - Silver Wizard design system, consistent with all SW2 apps

Use Cases

Multi-Agent Development Workflows

Development teams using multiple specialized AI agents across a codebase. Each agent has its own expertise (frontend, backend, infrastructure, testing).

Autonomous Overnight Development

Start a task, let agents coordinate overnight. They self-organize through messaging, spawn new agents as needed, reincarnate when contexts fill.

CI/CD-Like Agent Pipelines

One agent’s output triggers another agent’s task. Build agent finishes → Test agent spawns → Deploy agent takes over. All automatic.

Multi-Repo Projects

Different agents own different modules. Changes in one repo trigger coordination with agents managing dependent repos.

Agent Fleet Health Monitoring

Visual dashboard of context usage, message flow, active tasks. Spot bottlenecks, intervene when agents stall.

Architecture

┌──────────────────────────────────────────┐
│          IATV Control Center              │
│                                           │
│  ┌──────────┐  ┌──────────┐  ┌─────────┐│
│  │ Agent EE │  │ Agent MM │  │ Agent   ││
│  │ Terminal │  │ Terminal │  │ MacR    ││
│  │  [70%]   │  │  [45%]   │  │ [NOT    ││
│  └────┬─────┘  └────┬─────┘  │ RUNNING]││
│       │             │         └────┬────┘│
│       └─────────────┴──────────────┘     │
│                     │                     │
│            ┌────────▼────────┐            │
│            │  gmsg Broker    │            │
│            │  Message Queue  │            │
│            └────────┬────────┘            │
│                     │                     │
│       ┌─────────────┴───────────┐        │
│       │  Message arrives for    │        │
│       │  MacR → Auto-spawn!     │        │
│       └─────────────────────────┘        │
└──────────────────────────────────────────┘

What Makes It Different

vs. Manual Terminal Management:

  • Visual dashboard - See all agents at once, not scattered terminal windows
  • One-click launch - No remembering commands or paths
  • Auto-spawn - Agents wake each other up automatically

vs. tmux/screen:

  • Agent-aware - Understands Claude Code agents, not just shell sessions
  • Message-driven - Agents communicate through structured protocols
  • AfterLife integration - Transparent context management and reincarnation

vs. Other Multi-Agent Tools:

  • Terminal-native - Agents run in real terminal windows, not simulated
  • Event-driven spawning - Agents appear when needed, not pre-configured
  • Production infrastructure - Not a research demo, daily driver for real work

Tech Stack

  • Python 3.13 - Core application
  • PyQt6 - UI framework
  • PyteTerminalWidget - Embedded terminal emulation
  • gmsg MCP Server - Message broker
  • CC_Mem - Persistent memory
  • Claude Code CLI - Agent runtime

Status

Beta — Daily driver for Silver Wizard Software’s multi-agent development workflow. Currently manages 10+ agent types across the SW2 ecosystem.

Internal Infrastructure: IATV is production infrastructure developed for Silver Wizard Software’s multi-agent workflows. Available for licensing to organizations building AI agent systems.

Agent Types (Examples)

EE (Enterprise Architect):

  • Infrastructure, architecture review, technical planning
  • Icon: 🏗️

MM (Multi-Media Manager):

  • Content creation, social media, marketing
  • Icon: 📱

MacR (Mac Retriever):

  • Desktop application development
  • Icon: 🐕

PubliKit:

  • Publication and distribution workflows
  • Icon: 📚

Brand_Manager:

  • Brand strategy, messaging, product marketing
  • Icon: 📣

Best Practices

  1. Configure agent roles clearly - Each agent has a defined purpose and expertise
  2. Monitor context usage - Watch for agents approaching reincarnation threshold
  3. Use gmsg liberally - Let agents coordinate through messages, not manual coordination
  4. Start with user-led mode - Understand agent behavior before full autonomy

Licensing

IATV is internal production infrastructure available for licensing to organizations building multi-agent AI systems.

What you get: Full source code access, deployment documentation, and integration support for your multi-agent workflows.

Contact: silver.wizard001@proton.me for licensing inquiries.

Documentation

Documentation coming soon for licensees:

  • Getting Started Guide
  • Agent Configuration
  • Auto-Spawn Rules
  • Integration with CONF
  • Window Management

Support