Project introduction

OpenClaw is an open-source self-hosted AI assistant platform that connects messaging apps to AI agents running on your own hardware. It is designed for developers and advanced users who want control over data, deployment, and routing.

• Official website: https://openclaw.ai
• Docs: https://docs.openclaw.ai
• GitHub: https://github.com/openclaw/openclaw

OpenClaw is fully open source. This guide covers the practical steps required to connect it to the Dli.li AI OpenAI-compatible endpoint.

Core features

Multi-channel integration

• Telegram, Discord, WhatsApp, iMessage, and plugin-based extensions
• Unified gateway process
• Voice interaction support
• Interactive canvas UI support

Self-hosting and security

• Runs on your own hardware
• Open-source and auditable
• Context and skills stay local

Agent capabilities

• Persistent background execution
• Scheduled tasks
• Session isolation
• Multi-agent routing
• Native tool calling

Before connecting Dli.li AI

Requirements

- Node.js 22 or later - A working compatible endpoint; the official default API is `https://api.dli.li/v1`, and if you want the official backup API directly, use `https://api.dlizz.com/v1` - A working API key

It is recommended to get OpenClaw itself running first, then switch the model provider to Dli.li AI afterward. That makes troubleshooting much easier.

1. Install OpenClaw

curl -fsSL https://openclaw.ai/install.sh | bash

See the official Getting Started page for other install options.

2. Run onboarding

openclaw onboard --install-daemon

3. Verify gateway and dashboard

openclaw gateway status

openclaw dashboard

4. Find the config file

OpenClaw typically stores its configuration at ~/.openclaw/openclaw.json.

Useful environment variables

You can customize paths with:

• OPENCLAW_HOME
• OPENCLAW_STATE_DIR
• OPENCLAW_CONFIG_PATH

Use Dli.li AI as the provider

OpenClaw supports custom or OpenAI-compatible providers through models.providers. For Dli.li AI, the typical approach is to add a newapi provider and point default models to newapi/model-id.

Integration outline

1. Declare a newapi provider in models.providers.
2. Point baseUrl to your endpoint, including /v1.
3. Set api to openai-completions.
4. List the models you want to expose.
5. Point the default model to newapi/....

Recommended: store the key in an environment variable

export NEWAPI_API_KEY="sk-your-newapi-key"

Then update openclaw.json:

{
  models: {
    mode: "merge",
    providers: {
      newapi: {
        baseUrl: "https://api.dli.li/v1",
        apiKey: "${NEWAPI_API_KEY}",
        api: "openai-completions",
        models: [
          { id: "gemini-2.5-flash", name: "Gemini 2.5 Flash" },
          { id: "kimi-k2.5", name: "Kimi K2.5" }
        ]
      }
    }
  },
  agents: {
    defaults: {
      model: {
        primary: "newapi/gemini-2.5-flash",
        fallbacks: ["newapi/kimi-k2.5"]
      },
      models: {
        "newapi/gemini-2.5-flash": { alias: "flash" },
        "newapi/kimi-k2.5": { alias: "kimi" }
      }
    }
  }
}

Key fields

Field	Meaning
models.mode	Use merge to keep built-in providers and add newapi
models.providers.newapi.baseUrl	Your compatible endpoint, usually ending with /v1; the official default API is https://api.dli.li/v1, and the official backup API is https://api.dlizz.com/v1
models.providers.newapi.apiKey	Your API key, ideally injected from ${NEWAPI_API_KEY}
models.providers.newapi.api	Use openai-completions for a Dli.li AI compatible gateway
models.providers.newapi.models	Model IDs must match the site-exposed names
agents.defaults.model.primary	Default model, formatted as provider/model-id
agents.defaults.model.fallbacks	Fallback models
agents.defaults.models	Optional aliases for easier reference

Verify the result

Open the dashboard again:

openclaw dashboard

Or list models:

openclaw models list

If you can start a chat and see the newapi/ models in the list, the integration is working.

Common issues

• Missing /v1 in baseUrl
• Wrong model IDs
• API key only exported in the current shell while the gateway runs as a background service
• Need foreground logs for debugging: openclaw gateway --port 18789