skills.rest
OpenSourceSam
OpenSourceSamCOMMUNITY
Explore all skills by this author →
GitHub

token-efficient-delegation

Delegate LLM tasks to token-efficient models using a decision tree.

Updated Dec 15, 2025
Search Tags:#prompt engineering#llm#agent delegation#token efficiency#cost optimization#model selection
One-click install
npx skills add https://github.com/OpenSourceSam/v2_heras_garden --skill token-efficient-delegation
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: token-efficient-delegation
Source: https://github.com/OpenSourceSam/v2_heras_garden/tree/main/.claude/skills/token-efficient-delegation
Command: npx skills add https://github.com/OpenSourceSam/v2_heras_garden --skill token-efficient-delegation

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill addresses the high cost and latency associated with using powerful LLMs like Claude by providing a strategic framework for delegating tasks to more cost-effective and faster models when appropriate.

Core Features & Use Cases

  • Model Selection Guidance: Offers a decision tree and detailed model characteristics to guide task delegation.
  • Cost & Speed Optimization Patterns: Implements strategies like parallel execution, background processing, and cascade routing to minimize token usage and execution time.
  • Use Case: When needing to research a broad topic, instead of using an expensive Claude model, delegate the research to a cheaper MiniMax or GLM agent, then have Claude synthesize the findings, saving significant costs and time.

Quick Start

Delegate web research to a MiniMax agent for faster and cheaper results.

Frequently Asked Questions about token-efficient-delegation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I reduce LLM token usage and cost in agent workflows?▼

Reduce LLM token cost by delegating simple tasks to cheaper models like MiniMax or GLM based on a decision tree, reserving expensive models like Claude for synthesizing complex findings.

What is the best way to route LLM tasks for cost and speed optimization?▼

The best way to optimize LLM tasks is using cascade routing, parallel execution, and background processing patterns to delegate tasks to token-efficient models based on specific task types.

Can I use cheaper models for web research and Claude for synthesis?▼

Yes, you can delegate broad web research to a cheaper MiniMax or GLM agent for faster results, then use Claude to synthesize the findings, saving significant cost and time.

When should I use background execution for LLM operations?▼

Use background execution for long-running LLM operations to minimize execution time and latency, allowing agent workflows to continue processing other tasks while waiting for results.

What are the limitations of delegating tasks to token-efficient models?▼

Delegating to token-efficient models risks lower quality for complex reasoning tasks, so the framework enforces rules against using cheap models for tasks requiring high cognitive capability.

Related Skills

Explore More Software Engineering Skills→
michael-h-patriannamichael-h-patrianna

agent-selection

Select specialized agents for tasks using decision-tree logic.

Community
Advanced
takzobyetakzobye

flutter-concurrency

Runs Flutter JSON parsing and heavy tasks in background isolates.

Community
Advanced
tianzecntianzecn
2

multi-agent-coordination

Coordinate multiple agents in parallel or sequential workflows with delegation.

Community
Advanced
tranvietanh0tranvietanh0

t1k:context

Monitor context usage and optimize token consumption for LLM tasks.

Community
Advanced