Skip to main content

OpenClaw

OpenClaw is a self-hosted gateway that connects chat apps and channel surfaces to AI agents. Official site: docs.openclaw.ai

1.1 Usage​

  1. Run openclaw onboard for guided setup, or edit ~/.openclaw/openclaw.json directly.

  2. If your installation uses a custom location, check it with openclaw config file.

  3. Add a custom provider under models.providers.

  4. Set the provider to use the OpenAI-compatible transport:

    • api: openai-completions

    • baseUrl: https://api.hpc-ai.com/inference/v1

  5. Point the default model at the provider/model pair.

  6. Set apiKey to your HPC-AI key, or use OpenClaw's documented SecretRef or env-substitution flow.

Example Config​

{
  models: {
    providers: {
      hpc_ai: {
        baseUrl: "https://api.hpc-ai.com/inference/v1",
        api: "openai-completions",
        apiKey: "your-hpc-ai-api-key",
        models: [
          {
            id: "minimax/minimax-m2.5",
            name: "HPC-AI Minimax M2.5",
            reasoning: false,
            input: ["text"],
            contextWindow: 128000,
            maxTokens: 8192
          }
        ]
      }
    }
  },
  agents: {
    defaults: {
      model: { primary: "hpc_ai/minimax/minimax-m2.5" }
    }
  }
}

Method 2: Non-Interactive Onboarding​

export CUSTOM_API_KEY="your-hpc-ai-api-key"

openclaw onboard --non-interactive \
  --auth-choice custom-api-key \
  --custom-base-url "https://api.hpc-ai.com/inference/v1" \
  --custom-model-id "minimax/minimax-m2.5" \
  --custom-api-key "$CUSTOM_API_KEY" \
  --secret-input-mode plaintext \
  --custom-compatibility openai

1.2 References​