model-router

Classify tasks and select a Claude model tier before heavy work.

5|Updated Apr 8, 2026
One-click install
npx skills add https://github.com/JansenAnalytics/claudex --skill model-router-jansenanalytics
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: model-router
Source: https://github.com/JansenAnalytics/claudex/tree/main/skills/model-router
Command: npx skills add https://github.com/JansenAnalytics/claudex --skill model-router-jansenanalytics

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Quickly choosing the right Claude model tier before heavy AI work to balance cost and performance.

Core Features & Use Cases

  • Classifies a task and outputs an appropriate model tier (haiku/sonnet/opus).
  • Integrates with sub-agent workflows to optimize resource usage and speed.
  • Use Case: Before starting a long analysis or coding task, determine the optimal model tier to use for the task.

Quick Start

Spawn a sub-agent with the recommended tier after classifying the task description.

Frequently Asked Questions about model-router

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

FAQPage Schema
How do I choose the right Claude model tier for a coding or research task?

Classifying the task description evaluates complexity to output an appropriate Claude model tier—haiku, sonnet, or opus—balancing cost and performance before starting heavy analysis or coding.

What is model routing for sub-agent workflows?

Model routing for sub-agent workflows is the process of classifying a task first, then spawning the sub-agent with the recommended tier to optimize resource usage and speed.

How do I optimize AI costs when spawning sub-agents for heavy analysis?

You optimize AI costs by routing tasks to the appropriate tier before spawning sub-agents, ensuring lighter tasks use haiku while reserving opus for heavy analysis.

Do I need a specific environment setup to run the model classifier scripts?

You need access to a local MODEL_ROUTER_HOME directory, which defaults to ~/projects/model-router, to run the classify.cjs scripts that determine the tier.

When should I not use an automated model tier selector?

You should skip automated tier selection for trivial tasks where the overhead of running classifier scripts outweighs the cost savings, or if the task context falls outside coding, research, and decision workflows.