Verified:GitHub

intelli-router

Classify user messages by complexity and route them to AI model tiers.

4|Updated Feb 6, 2026

Search Tags:#agent workflow #model routing #ollama #llm triage #api cost optimization #prompt classification

One-click install

npx skills add https://github.com/ColeMatthewBienek/IntelliRouter --skill intelli-router

Or copy as Structured Prompt for Agent▼

Please help me install this Agent Skill.
Skill: intelli-router
Source: https://github.com/ColeMatthewBienek/IntelliRouter/tree/main
Command: npx skills add https://github.com/ColeMatthewBienek/IntelliRouter --skill intelli-router

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires json, sys, re, requests, subprocess, os, time, and includes scripts (resource) components.

What problem does it solve?

IntelliRouter reduces AI API costs and latency by automatically classifying incoming requests by complexity and routing each one to the most appropriate model tier.

Core Features & Use Cases

Automated complexity triage: Uses a local Ollama model to classify each incoming user message into simple, moderate, coding, or complex tiers.
Cost-aware model routing: Routes simple tasks to a local model and escalates to higher-capability API models only when needed.
Always-on instruction for OpenClaw: Designed to run on every user message (except slash commands) before the agent generates a response, ensuring consistent routing behavior.

Use case: An agent handles a mixed chat where some messages are greetings or quick questions while others request debugging or architecture-level analysis; this skill keeps simple traffic local and reserves expensive models for the truly complex prompts.

Quick Start

Tell your OpenClaw agent to read the skill file at ~/.openclaw/workspace/skills/intelli-router/SKILL.md and use it as a skill.

Frequently Asked Questions about intelli-router

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

FAQPage Schema

How does model routing reduce API costs for LLM agents?▼

Model routing lowers API costs by classifying prompt complexity and sending simple tasks to local models, reserving expensive API models for complex prompts that require higher capability.

How do I set up LLM triage to route prompts using Ollama?▼

To set up LLM triage, configure an existing Ollama endpoint with a local triage model, then instruct your OpenClaw agent to read the skill file to classify messages into routing tiers via HTTP requests.

Do I need an Ollama endpoint to use automated prompt classification?▼

Yes, an existing Ollama endpoint and local triage model are required, as the skill sends HTTP requests to classify prompts and determine the appropriate model tier before routing.

What happens when prompt classification fails during model routing?▼

When prompt classification fails, the skill normalizes invalid tier responses and automatically falls back to the moderate tier, ensuring the agent still generates a response without interruption.