A unified LLM gateway providing a single OpenAI-compatible API to access over 70 different AI models from one interface.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-aisa-llm-gateway
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 openclaw-aisa-llm-gateway 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 LLM Router?
The LLM Router is a professional-grade gateway designed for autonomous agents and developers who need to consolidate their AI infrastructure. By replacing over 100 individual API keys with a single point of access, this skill allows for seamless switching between models like GPT-4, Claude 3.5, Gemini 2.0, Qwen, and Deepseek. It is built to enhance the capabilities of Openclaw Skills by providing a robust, OpenAI-compatible interface that simplifies multi-model integration.
Beyond simple access, the router provides a standardized layer for billing, cost tracking, and response formatting. This allows developers to focus on building intelligent workflows rather than managing fragmented API documentation or fluctuating provider-specific SDKs. It is the definitive solution for those looking to leverage the full spectrum of modern LLMs within a single, unified environment.
LLM Router Use Cases
- Implement cost-optimized routing by directing simple queries to inexpensive models and complex tasks to high-end reasoning engines.
- Create reliable fallback strategies that automatically switch to a secondary provider if the primary model encounters an error or rate limit.
- Conduct model A/B testing and comparison to evaluate which AI provides the best response for specific specialized tasks.
- Execute vision and multimodal analysis across different model families using a single standardized request format.
- Select specialized models for niche tasks such as Deepseek for coding or Qwen for multilingual support within Openclaw Skills.
How LLM Router Works
- The user authenticates using a single AISA_API_KEY environment variable.
- A request is sent to the unified endpoint specifying the target model name in the JSON payload.
- The router interceptor identifies the provider and translates the request into the native format required by the specific AI model.
- The response is normalized back into an OpenAI-compatible structure, including unified usage and cost metadata.
- The integrated client allows for either synchronous completion or real-time streaming via Server-Sent Events.
LLM Router Setup
To begin using the LLM Router within your Openclaw Skills environment, configure your API key as follows:
export AISA_API_KEY="your-key"
You can then verify the setup by listing available models using the provided Python utility:
python3 scripts/llm_router_client.py models
LLM Router Data Schema & Taxonomy
The skill organizes its interactions using a strictly OpenAI-compatible schema to ensure maximum interoperability.
| Attribute | Type | Description |
|---|---|---|
| model | string | The specific model identifier (e.g., gpt-4o, claude-3-opus) |
| messages | array | A list of role-based message objects (system, user, assistant) |
| usage.cost | float | The calculated cost of the transaction in USD |
| usage.total_tokens | integer | The sum of prompt and completion tokens used |
All metadata, including remaining credits and per-model pricing, is returned in the response headers or the usage object of the JSON response.
LLM Router Advanced Features
- Support for multi-modal vision analysis allowing image data to be processed by any vision-capable model.
- Native function calling and tool definitions to enable agents to interact with external APIs structuredly.
- Built-in streaming support using Server-Sent Events (SSE) for high-performance, low-latency user interfaces.
- Automatic response caching and latency monitoring for optimized agent performance.
- Comprehensive Python SDK that operates using only standard libraries, ensuring zero-dependency integration for Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Media Gen
A unified tool for generating cinematic images and videos using Gemini 3 Pro and Qwen Wan 2.6 via a single API key.
MarketPulse
MarketPulse is a comprehensive financial data skill providing real-time and historical equities and cryptocurrency data for autonomous agents.
MarketPulse
A high-performance financial data skill for AI agents to query real-time and historical equity and cryptocurrency markets.
AIsa Provider
A unified API gateway providing production access to China's leading AI models with official partnerships and enterprise-level data privacy.
OpenClaw Search
An intelligent multi-source search skill for agents that integrates web, academic, and Tavily retrieval with automated confidence scoring.
US Stock Analyst
A professional equity analysis skill that generates comprehensive market reports using real-time data and multi-model AI.