An intelligent model dispatching system that uses multi-dimensional analysis to route requests to the most efficient AI model based on complexity and cost.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install smart-router-skill
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-router-skill 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 Intent Router Skill?
The Smart Intent Router is a sophisticated model orchestration layer designed to transcend simple keyword matching. By leveraging multi-dimensional analysis—including message length, content complexity, and pattern recognition—it ensures every request is handled by the optimal LLM.
As a critical addition to your collection of Openclaw Skills, this system integrates cost-aware logic to prioritize free models for simple tasks while reserving premium models for high-complexity requirements. It provides a robust framework for managing multiple AI providers like OpenRouter and DeepSeek, effectively balancing high-performance output with operational expenditure.
Smart Intent Router Skill Use Cases
- Automatically routing coding queries to specialized high-reasoning models like Claude 3.5 Sonnet.
- Offloading simple conversational tasks or greetings to free models on OpenRouter to preserve budget.
- Enforcing strict daily or monthly API spend limits through automated budget controls and smart alerts.
- Managing complex multi-agent workflows where different tasks require specific model capabilities.
- Real-time performance monitoring and health checks for high-availability AI applications.
How Smart Intent Router Skill Works
- The system performs a multi-dimensional analysis of the incoming message, evaluating length, complexity, and specific patterns.
- It identifies the intent and context of the dialogue to determine the necessary reasoning capabilities.
- The router checks the configured priority rules and cost-optimization settings in smart-router-config.json to select the best agent.
- It executes the request through the chosen model while monitoring performance metrics and tracking cost data in real-time.
- Automated maintenance tasks periodically scan for new free models and optimize routing rules based on historical usage data.
Smart Intent Router Skill Setup
To install this skill, ensure you have OpenClaw 2026.2.13 or higher and Node.js 18+ installed.
# 1. Copy the skill to your OpenClaw directory
cp -r smart-router-skill /opt/homebrew/lib/node_modules/openclaw/skills/
# 2. Enable the skill in your configuration
openclaw config patch --json '{"skills":{"entries":{"smart-router":{"enabled":true}}}}'
# 3. Set your API keys
export OPENROUTER_API_KEY="your-openrouter-api-key"
export DEEPSEEK_API_KEY="your-deepseek-api-key"
# 4. Initialize and restart the gateway
node enhanced-intent-router.js init
openclaw gateway restart
Smart Intent Router Skill Data Schema & Taxonomy
The skill manages its logic through a structured JSON configuration and maintains metadata for performance tracking.
| Component | Configuration Key | Description |
|---|---|---|
| Routing Rules | routingRules.priority |
Array of conditions including keywords, minLength, and target agent. |
| Fallback | routingRules.fallback |
The default model used when no specific rules are triggered. |
| Cost Control | costOptimization |
Defines budget limits, daily caps, and free model preferences. |
| Performance | performanceSettings |
Manages cache settings (TTL) and maximum allowed response times. |
Smart Intent Router Skill Advanced Features
- Automated OpenRouter scanning to discover and integrate new free models every 24 hours.
- Dynamic rule optimization that adjusts routing priorities based on actual performance and cost logs.
- Intelligent caching with adjustable TTL to reduce latency and redundant analysis for similar intents.
- Multi-agent support enabling seamless transitions between code, research, and creative agents.
- Automated backup and recovery system for all routing configurations and performance reports.
- Hot-reloading capabilities allowing rule updates without restarting the OpenClaw service.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Reference Manual
A complete technical reference for managing OpenClaw configurations, CLI operations, and multi-channel AI agent deployments.
DocuClaw
A sovereign document intelligence and archival system that extracts structured data from files 100% locally using AI.
DeepReader
A high-performance web and social media reader that converts URLs into AI-ready Markdown without requiring external API keys.
DeepBook CLI Skill
A powerful command-line interface for end-to-end DeepBook operations including market data, account management, and on-chain execution.
Skill Search
A multi-modal search engine for AI agents to discover, match, and load specialized capabilities through keyword, semantic, or LLM-powered matching.
record CLI
A powerful macOS command-line utility for capturing audio, screen recordings, screenshots, and camera input, optimized for terminal-based AI agents.
Featured*
