Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Butler manages token budgets across multiple AI API providers, spawns sub-agents for parallel task execution, and rotates API keys automatically at 75% usage. It tracks 8 API keys across 6 providers with a combined capacity of 28.5M tokens per day. Task decomposition splits complex prompts into prioritized sub-tasks with proportional budget allocation.
Centralizes multi-provider token management and concurrent agent orchestration in one library instead of tracking usage across separate provider dashboards and writing custom retry logic.
When to use it
- Running parallel sub-agents to research and summarize from multiple sources
- Automatically switching API keys mid-task to avoid hitting rate limits
- Splitting a data pipeline into concurrent steps with per-stage token caps
- Monitoring daily token spend across Anthropic, OpenAI, and Groq from one status call
- Retrying failed sub-tasks without restarting the full agent workflow
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Research AI agent frameworks, analyze their capabilities, and write a comparison report
- 1Reads PRD file to estimate token complexity and selects an appropriate API key
- 2Decomposes the task into sub-tasks: research (30% budget), analyze (40%), write report (30%)
- 3Spawns up to 3 concurrent sub-agents with proportional token allocations
- 4Monitors usage per sub-agent and rotates to a new key if any session crosses 75% of its allocation
- 5Aggregates results and returns success rate, tokens consumed per sub-task, and combined output
Aggregated comparison report with per-subtask status, total tokens used, and any retry details
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.