higress-openclaw-integration

Deploy Higress AI Gateway and configure OpenClaw as a unified LLM provider.

9.3k|1.3k|Updated Oct 27, 2022
One-click install
npx skills add https://github.com/alibaba/higress --skill higress-openclaw-integration
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: higress-openclaw-integration
Source: https://github.com/alibaba/higress/tree/main/.agents/skills/higress-openclaw-integration
Command: npx skills add https://github.com/alibaba/higress --skill higress-openclaw-integration

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Managing multiple LLM provider API keys and endpoints is fragmented and error-prone. This Skill deploys Higress AI Gateway as a single local endpoint and connects OpenClaw to it, so all model providers are routed through one gateway with hot-reloadable configuration.

Core Features & Use Cases

  • Gateway Deployment: Downloads and runs the Higress AI Gateway install script with non-interactive flags for 20+ providers (z.ai, Claude, OpenAI, DeepSeek, Moonshot, Dashscope, and more).
  • Region-Aware Configuration: Detects China vs. international timezone via a script to select the correct z.ai endpoint domain (open.bigmodel.cn or api.z.ai).
  • OpenClaw Plugin Installation: Copies a TypeScript provider plugin into the OpenClaw extensions directory, enabling models with the higress/ prefix and auto-routing via higress/auto.
  • Hot-Reload Management: Adds or updates API keys, routing rules, and provider domains without restarting the gateway container.
  • Use Case: A developer wants OpenClaw to use glm-5 for coding and Claude for reasoning. This Skill deploys the gateway with both keys, installs the plugin, and configures auto-routing so requests are routed by message content.

Quick Start

Ask the AI to deploy Higress AI Gateway with your provider API key and connect it to OpenClaw with auto-routing enabled.

Frequently Asked Questions about higress-openclaw-integration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I deploy Higress AI Gateway for OpenClaw?

Run the get-ai-gateway.sh install script with the --non-interactive flag and your provider key, such as --zhipuai-key or --openai-key. Then copy the plugin files to ~/.openclaw/extensions/higress and run openclaw plugins enable higress followed by openclaw models auth login --provider higress.

Which LLM providers does Higress AI Gateway support?

The gateway supports over 20 providers including z.ai, Claude, OpenAI, DeepSeek, Moonshot, Minimax, Dashscope, Grok, Gemini, OpenRouter, Groq, Doubao, Mistral, and cloud providers like Azure OpenAI, AWS Bedrock, and Google Vertex AI.

How does auto-routing work with higress/auto?

Enable auto-routing at deployment with --auto-routing and a default model. Requests using model higress/auto are routed based on message content, and you can add rules later with get-ai-gateway.sh route add --model <model> --trigger "keyword1|keyword2".

Can I update API keys without restarting the gateway?

Yes, API key changes are hot-reloaded. Run get-ai-gateway.sh config add --provider <provider> --key <api-key> and the change takes effect immediately without restarting the container or OpenClaw.

Why does the gateway container fail with too many open files?

The system's fs.inotify.max_user_instances limit is too low, often 128 by default. Increase it to 8192 with sudo sysctl -w fs.inotify.max_user_instances=8192, persist it in /etc/sysctl.conf, then restart the container.

How is the z.ai endpoint domain selected for China vs international users?

A detection script reads the system timezone and returns china or international. China users get open.bigmodel.cn by default, while international users get api.z.ai added automatically via the --zhipuai-domain flag.