A sophisticated routing skill that automatically selects the most efficient AI model from your existing local pool based on task complexity and data type.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install universal-model-switcher
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install universal-model-switcher using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Universal Model Switcher?
The Universal Model Switcher is a high-performance optimization skill designed to streamline how AI agents interact with multiple Large Language Models. Built for the Openclaw Skills ecosystem, this tool operates on the core principle of maximizing your existing resources without incurring extra costs. It intelligently categorizes tasks—ranging from complex code generation to long-form document analysis—and assigns them to the most capable model in your local inventory.
By implementing a multimodal-first approach and specialized task-based switching rules, the skill ensures that high-reasoning tasks go to powerful models like qwen3-max, while coding assignments are handled by experts like glm-5. This results in faster response times and higher output quality across all your automated workflows.
Universal Model Switcher Use Cases
- Automatically routing programming tasks to specialized coding models to improve syntax accuracy.
- Handling massive datasets and long documents using high-context models like qwen3.5-plus.
- Delegating complex logical reasoning and market analysis to dedicated subagent profiles.
- Streamlining multimodal interactions involving images, video, and audio through optimized processing paths.
- Maintaining high-quality office productivity workflows using specialized models like MiniMax-M2.5.
How Universal Model Switcher Works
- The skill intercepts the incoming prompt and identifies the nature of the task (e.g., multimodal, coding, or reasoning).
- It references a predefined model pool to match the task requirements with the primary 'Best' model available.
- If the request originates from a specific subagent, such as a Tech Expert, the skill applies specialized professional routing rules.
- The system attempts to initiate the session with the primary model; if unavailable, it seamlessly switches to the designated 'Backup' model.
- The task is executed within the chosen model's parameters, ensuring optimal performance and cost-efficiency within the Openclaw Skills framework.
Universal Model Switcher Setup
To deploy the Universal Model Switcher, ensure your local environment has the necessary models pulled and ready for invocation. Follow these steps to configure your Openclaw Skills environment:
# Initialize the skill configuration
openclaw skills install universal-model-switcher
# Verify your local model pool availability
ollama list | grep -E 'glm-5|qwen3|minimax'
# Set the skill as the primary router for your agent workflows
openclaw configure --router universal-model-switcher
Universal Model Switcher Data Schema & Taxonomy
The skill organizes routing logic based on a structured hierarchy of tasks and subagent roles. Below is the primary data mapping used for model selection:
| Task Type | Primary Model | Backup Option | Specialized Feature |
|---|---|---|---|
| Coding | glm-5 | qwen3-coder-plus | Syntax Optimization |
| Reasoning | qwen3-max | qwen3.5-397b-a17b | Complex Logic |
| Multimodal | qwen3.5-plus | N/A | Vision/Audio Support |
| Long Doc | qwen3.5-plus | N/A | 256K Context Window |
| Office | MiniMax-M2.5 | N/A | Productivity Tasks |
Universal Model Switcher Advanced Features
- Professional Subagent Routing: Automatically assigns specialized models to roles like Tech Expert, Reasoning Expert, and Market Hunter.
- Multimodal First Logic: Prioritizes qwen3.5-plus for any input containing non-text media for superior analysis.
- Context-Aware Switching: Intelligent selection based on document length, utilizing models with up to 256K token support.
- Zero-Cost Optimization: Strictly limits switching to models already present in your local pool to avoid unexpected API fees.
- Robust Fallback System: Ensures continuity by falling back to high-performing secondary models if the primary choice is busy or unavailable.
SKILL.md
Loading
Related Openclaw Skills
writER
A sophisticated writing style clone designed to transform casual drafts and oral inputs into poetic, philosophical prose and romantic literature.
Macro News Signal
An intelligent market analysis skill that transforms real-time global news and key macro indicators into structured investment insights.
WeChat Typeset Pro
A professional typesetting engine that converts Markdown articles into exquisitely styled, WeChat-compatible HTML with over 30 customizable themes and AI-driven content enhancement.
Instant Client Audit Report Generator
A high-performance agency tool that automates comprehensive prospect audits and sales pitches using real-time data and AI.
Meeting Minutes Fusion Generator
A high-performance tool that merges transcriptions from DingTalk and Get into a structured, professionally designed 10-block meeting report in HTML and PDF formats.
MindStudio Generate Text Block Prompting Skill
A specialized skill for crafting production-ready prompts within MindStudio workflows, ensuring reliable text and structured JSON outputs.
Featured*
