AgentPod
AgentPod provides remote AI development workstations with integrated terminal access. Each Pod runs on a Runner and provides a full development environment with AI assistance.
Overview
An AgentPod is an isolated development environment where an AI coding agent (like Claude Code) runs with full terminal access. Each Pod includes:
- Web Terminal — Real-time terminal access via WebSocket
- Git Worktree — Isolated workspace per Pod
- AI Agent — Your chosen coding assistant
- MCP Tools — 24 tools for discovery, terminal control, collaboration, and more
Key Features
Web-based Terminal
Full terminal emulation in your browser with WebSocket support for real-time interaction. No local setup required.
Agent Status Monitoring
Real-time visibility into what your AI agent is doing: idle, working, waiting for input, or finished.
Git Worktree Isolation
Each Pod gets its own Git worktree, ensuring complete isolation between concurrent Pods.
Multiple Pods
Run multiple AI agent Pods concurrently, each working on different tasks or branches.
Ticket Integration
Link Pods to tickets for automatic context and progress tracking.
Automatic Cleanup
Pods are automatically cleaned up when terminated, freeing resources.
Pod Lifecycle
Pod is being created, worktree is being set up, and agent is starting.
Pod is active and ready for interaction.
Pod is temporarily paused (agent not active).
Runner lost connection but Pod may recover.
Pod finished successfully.
Pod was manually stopped.
Pod lost its Runner connection and cannot recover.
Pod encountered an error during execution.
Agent Status
Within a running Pod, the AI agent has its own status:
| Status | Description |
|---|---|
| Idle | Agent is waiting for a task or prompt |
| Executing | Agent is actively executing a task |
| Waiting | Agent is waiting for user input or approval |
Pod Configuration
When creating a Pod, you can configure:
| Option | Description |
|---|---|
| Agent Type | Claude Code, Codex CLI, Gemini CLI, Aider, or custom |
| Model | opus, sonnet, or haiku (for Claude) |
| Permission Mode | plan (ask before changes), default, or bypassPermissions |
| Repository | Link to a Git repository for automatic cloning |
| Ticket | Link to a ticket for context and tracking |
| Initial Prompt | Starting instructions for the AI agent |