An intelligent routing layer that selects the most cost-effective AI model for any given task while ensuring zero downtime from rate limits.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install chitin-core
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 chitin-core 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 Chitin Core?
Chitin Core is a vital component for managing multi-model environments within Openclaw Skills. It acts as an abstraction layer between the user's intent and the underlying LLM providers. By analyzing task complexity, it automatically assigns work to the appropriate tier—ranging from lightweight, low-cost models for simple queries to heavy-duty reasoning models for complex architecture.
This skill ensures that you never overpay for simple tasks or underpower critical reasoning processes. It is designed to maximize the efficiency of Openclaw Skills by providing a centralized model registry and a logic-driven selection process that considers cost, capability, and current provider health.
Chitin Core Use Cases
- Spawning specialized sub-agents for modular coding tasks without manual model selection.
- Reducing total API costs by delegating simple Q&A and status checks to cheaper models.
- Implementing high-availability AI systems that automatically skip rate-limited providers.
- Testing different model tiers using override tags for performance and latency comparison.
How Chitin Core Works
- The user or a master agent provides a task description to the router script.
- Chitin Core analyzes task signals to classify the requirement into LIGHT, MEDIUM, or HEAVY tiers.
- The router selects the most cost-effective model currently available within that tier.
- If a spawned session fails due to a rate limit or error, the skill logs the failure and excludes that model from immediate retries.
- The system performs graceful degradation, automatically escalating to more capable tiers or falling back to local instances to maintain service uptime.
Chitin Core Setup
To set up Chitin Core, ensure it is installed in your workspace and configure your available models:
- Navigate to the skill directory:
cd ~/.openclaw/workspace/skills/chitin-core
- Edit the
config.jsonfile to include your API keys, model identifiers, and specific cost figures. - Validate your configuration to ensure all endpoints are reachable:
node scripts/router.js validate
Chitin Core Data Schema & Taxonomy
The router organizes data based on task complexity and model performance. Model selection responses are returned in a structured JSON format to be consumed by other Openclaw Skills components:
| Attribute | Description |
|---|---|
| tier | The classified level (LIGHT, MEDIUM, HEAVY) |
| model | The specific provider/model string for API calls |
| confidence | A decimal representing the routing engine's certainty |
| estimatedCost | Predicted cost of the transaction based on config.json |
| signals | Reasoning markers used for the classification decision |
Chitin Core Advanced Features
- Dynamic tier overrides: Force specific tiers by adding @light, @medium, or @heavy tags directly to your task text.
- Local fallback support: Automatically shift workloads to local Ollama instances if cloud providers are offline or rate-limited.
- Health & Cost Monitoring: Built-in commands to view real-time provider health status and accumulated usage costs.
- Intelligent Retry Logic: A state-aware failure handling system that prevents 'retry loops' on broken endpoints by skipping failed models during re-routing.
SKILL.md
Loading
Related Openclaw Skills
Chitin Editorial
A git-backed coordination system for AI agents to manage content publication, prevent duplicates, and track narrative timelines.
Deepgram Discord Voice
Enables real-time, low-latency voice conversations with AI agents in Discord channels using Deepgram streaming technology.
Cavos CLI Starknet Wallet Manager
A command-line interface skill for managing Starknet assets and smart contract interactions with session-based security.
Ressemble AI Voice Integration
A powerful integration for Openclaw Skills that provides professional-grade text-to-speech synthesis and speech-to-text transcription via Resemble AI.
Chitin Moat
A security framework that enforces AI agent permission boundaries based on the communication channel trust level.
Mobayilo Voice
A safety-first voice automation skill that enables AI agents to place outbound phone calls for bookings and follow-ups.
Featured*
