# Claude Code Orchestrator

> Spawn and manage sibling Claude Code agents through claude-agent for parallel frontend, workbench, implementation, and review tasks.

- Skill: `kingbootoshi/claude-code-orchestrator-2` (Agent Skill)
- Install (CLI): `npx skillmds@latest add kingbootoshi/claude-code-orchestrator-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/kingbootoshi/claude-code-orchestrator-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Web & Frontend
- Author: kingbootoshi (https://skillmd.com/u/kingbootoshi)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/kingbootoshi/claude-code-orchestrator-2

---


# Claude Code Orchestrator

Use the local `claude-agent` CLI to run background Claude Code agents in tmux:

```bash
claude-agent start "Task prompt" --dir /path/to/repo --map --xhigh
claude-agent await-turn <jobId>
claude-agent capture <jobId> 160 --clean
claude-agent send <jobId> "Continue with this correction."
```

Defaults are `--model opus --effort max --permission-mode auto`. Use `--xhigh` for broad workbench variant generation and default `max` for final review or hard implementation passes.

For workbench jobs, keep the prompt compact:

```text
Read apps/workbench/AGENT_DESIGN_RULES.md, src/generated/theme-summary.md, and the target brief.md.
Create v1-v5 distinct variants, run bun run check, and report files changed.
```

Use `claude-agent jobs --json` for compact status. Use `claude-agent attach <jobId>` for direct tmux interaction.