Skip to content

AI Agents

Shellarium runs AI coding agents as headless subprocesses. Each agent receives your task description and works autonomously to implement it.

  • Claude Code — Anthropic’s coding agent (default)
  • Codex CLI — OpenAI’s coding agent
  • ACP-compatible agents — any agent implementing the Agent Communication Protocol

When you start an agent, Shellarium:

  1. Creates an isolated git worktree for the workspace
  2. Spawns the agent process with your task as the prompt
  3. Streams the agent’s output to the UI in real time
  4. Handles permission requests from the agent

The agent has full access to the worktree and can read files, write code, run commands, and use tools — subject to your permission settings.

The output panel shows:

  • Thinking — the agent’s reasoning process
  • Tool calls — files read, code written, commands executed
  • Permission requests — actions requiring your approval
  • Cost tracking — token usage and estimated cost

If an agent is interrupted or you want to send follow-up instructions, Shellarium supports session resumption. The agent picks up where it left off with full context preserved.

You can choose which model to use for each workspace:

  • Claude Sonnet — fast and capable, good for most tasks
  • Claude Opus — most capable, best for complex tasks
  • Claude Haiku — fastest, good for simple tasks