OpenClaw Integration Guide

Use Step’s reasoning model in OpenClaw

Prerequisites

Before configuring OpenClaw, make sure you have:

• Subscribed to Step Plan
• Logged in to platform.stepfun.ai and created an API Key in the console.

Install OpenClaw

If you haven’t installed OpenClaw yet, complete the installation first.

Run one of the following commands in your terminal:

macOS / Linux / WSL2

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

Windows

powershell -c "irm https://openclaw.ai/install.ps1 | iex"

npm

npm i -g openclaw

Configure the Step Model

After installation, the Onboard wizard launches automatically. If you’ve already installed OpenClaw, you can re-launch the Onboard wizard with:

openclaw onboard

Configure the model API key

Use the following settings:

• In Model/auth provider, select StepFun.
• In StepFun auth method, select StepFun Step Plan API key.
• In Enter StepFun API key, paste your API key.
• Model configured displays the default model under this auth method.
• In the Default model list, you can keep the default or pick another model.

Complete the remaining configuration

Next, configure channel, search, skills, and hooks based on your needs. After configuration, you can choose to launch in the terminal (TUI) or in the Web UI.

Verify

After launching, ask any question and check whether OpenClaw responds in the interface.

Other configuration methods

Quick setup via official script

macOS / Linux (Bash)

curl -fsSL https://raw.githubusercontent.com/Zgh332358/openclaw-stepfun-installer/main/add_stepfun_smart.sh | bash

Windows (PowerShell, administrator mode)

Note: You must run this in PowerShell as administrator. Otherwise the script will print a notice and exit.

1. Right-click the Windows Start menu and choose “Terminal (Admin)” or “Windows PowerShell (Admin)”.

2. Run the following command:

irm https://raw.githubusercontent.com/Zgh332358/openclaw-stepfun-installer/main/add_stepfun_smart.ps1 | iex

Edit the configuration file manually

The OpenClaw configuration file is located at ~/.openclaw/openclaw.json.

Open it in your editor of choice and add or modify the following:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "stepfun/<model_id>" // <model_id> can be step-3.5-flash-2603 or step-3.5-flash
      }
    }
  },
  "models": {
    "mode": "merge",
    "providers": {
      "stepfun": {
        "baseUrl": "https://api.stepfun.ai/step_plan/v1",
        "apiKey": "${STEP_API_KEY}",
        "api": "openai-completions",
        "models": [
          {
            "id": "<model_id>",
            "name": "<model_id>",
            "reasoning": true,
            "contextWindow": 256000,
            "maxTokens": 8192
          }
        ]
      }
    }
  }
}

FAQ

Q: 'Model is not allowed' after configuration

If you’ve configured agents.defaults.models (an allowlist), make sure stepfun/<model_id> is also added to that list, otherwise the model will be blocked. To fix:

{
  "agents": {
    "defaults": {
      "models": {
        "stepfun/<model_id>": {
          "alias": "Step Model"
        }
      }
    }
  }
}

Or simply remove the agents.defaults.models entry to disable the allowlist.

Q: Why does the `api` field use `openai-completions`?

StepFun’s API is fully compatible with the OpenAI Chat Completions request format. When configuring a custom provider in OpenClaw, two API types are supported:

• openai-completions: for providers compatible with the OpenAI format (most third-party providers)
• anthropic-messages: for providers compatible with the Anthropic Messages format

StepFun belongs to the former.

Q: Can I skip WSL2 on Windows?

Not recommended. OpenClaw depends on Linux toolchains (make, g++, etc.) and systemd service management; running natively on Windows runs into many compatibility issues. WSL2 is the officially recommended way to use OpenClaw on Windows.

References

• OpenClaw Documentation
• OpenClaw Model Providers
• OpenClaw Models CLI
• Step Plan subscription