An intelligent multi-model router that automatically selects the optimal AI model for any task using @alias shortcuts or auto-classification.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install smart-models
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 smart-models 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 Smart Router?
The Smart Router is a sophisticated orchestration tool designed to enhance Openclaw Skills by providing a unified interface for over 35+ AI models across 7 distinct categories. Instead of being locked into a single provider or model, this skill intelligently evaluates incoming requests—whether they involve complex reasoning, image generation, or code debugging—and routes them to the most capable endpoint.
By leveraging any OpenAI-compatible API, the Smart Router ensures high availability and cost-efficiency. It acts as a middleware layer for your AI agent, allowing it to switch seamlessly between specialized models like GPT-4o for vision, Sora for video, or DeepSeek for reasoning without requiring manual configuration for every prompt.
Smart Router Use Cases
- Analyzing and describing images or performing OCR via vision-capable models.
- Generating high-fidelity images, videos, or audio assets based on text prompts.
- Solving complex mathematical proofs or logic puzzles requiring deep reasoning chains.
- Developing and debugging code using specialized programming LLMs.
- Streamlining daily productivity tasks with general-purpose chat models within Openclaw Skills.
How Smart Router Works
- The skill receives a prompt and checks for a specific @alias prefix, such as @o3 for reasoning or @flux for image generation.
- If no alias is present, the system applies auto-classification rules to detect the task type based on the content of the message.
- The router consults the models.json configuration to identify the default model for the detected category.
- It executes a bash script to call the selected model via a configured OpenAI-compatible API endpoint.
- In the event of a model call failure, the system automatically falls back to the next available model in the same category to ensure continuity.
Smart Router Setup
To integrate this into your Openclaw Skills environment, you must configure your API credentials and provider details.
Set the following environment variables in your terminal or configuration file:
export SMART_ROUTER_BASE_URL="https://api.openai.com/v1"
export SMART_ROUTER_API_KEY="your_api_key_here"
You can customize your model selection by editing the models.json file. To discover new models provided by your API, run the sync script:
scripts/sync-models.sh
Smart Router Data Schema & Taxonomy
The Smart Router manages its logic through a structured configuration file and environment variables.
| Data Component | Description |
|---|---|
| models.json | The primary configuration file defining categories, model IDs, and default flags. |
| @alias Table | A mapping of short commands to specific model IDs for instant manual routing. |
| API Meta | Uses OpenAI-compatible endpoints to standardize requests across different providers. |
| Task Types | Categorizes requests into chat, image, async, or tts for specific handling. |
Smart Router Advanced Features
- Intelligent auto-classification that detects intent for vision, coding, reasoning, and media tasks.
- Extensive @alias library supporting over 35+ specialized models for power-user control.
- Robust fallback logic that automatically retries failed requests using alternative models in the same category.
- Support for asynchronous workflows, enabling long-running video and audio generation tasks without blocking Openclaw Skills.
- Fully customizable model taxonomy via models.json, allowing for the addition of new providers and experimental LLMs.
SKILL.md
Loading
Related Openclaw Skills
Desktop Pet
An animated, always-on-top pixel art companion that reacts to your AI agent's activity and roams your macOS screen.
HERMES - AI Agent Evolution System
A next-generation AI agent knowledge sharing and evolutionary platform that moves beyond passive feedback to active prediction and decentralized skill management.
Mersoom AI Client
A specialized client for AI agents to engage in anonymous social networking, featuring automated Proof of Work solving and persistent memory.
x402 Client Agent Payment Skill
A protocol-driven payment skill that allows AI agents to exchange USDC stablecoins over HTTP for services and APIs.
WhereCanIWatch
A specialized tool for AI agents to locate movies and TV shows across all major US streaming platforms and digital storefronts.
nostr-nak
A robust CLI-based bridge for the Nostr protocol designed to work seamlessly with AI agents through Openclaw Skills.
Featured*
