OpenClaw

常用操作

1.首次添加模型

• 命令行添加

# 1.添加模型
openclaw onboard --non-interactive --accept-risk \
--auth-choice custom-api-key \
--custom-provider-id sensenova \
--custom-base-url https://token.sensenova.cn/v1 \
--custom-model-id sensenova-6.7-flash-lite \
--custom-api-key "sk-vMe50Kfc" \
--custom-compatibility openai \
--custom-image-input


# 2.设置为全局默认
openclaw models set sensenova/sensenova-6.7-flash-lite

# 3.再次list，看到 default 标签就代表成功
openclaw models list

# 查看模型列表
openclaw models list

• 配置文件添加

配置文件

{
  "models": {
    "providers": {
      "sensenova": {
        "baseUrl": "https://token.sensenova.cn/v1",
        "api": "openai-completions",
        "apiKey": "sk-vMe50Kfc",
        "models": [
          {
            "id": "sensenova-6.7-flash-lite",
            "name": "sensenova-6.7-flash-lite (Custom Provider)",
            "contextWindow": 128000,
            "maxTokens": 4096,
            "input": ["text", "image"],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "reasoning": false
          }
        ]
      }
    }
  }
}

1.1 追加模型

openclaw config set models.providers.sensenova.models \
'[
  {
    "id":"deepseek-v4-flash",
    "name":"DeepSeek V4 Flash（商汤代理｜1M上下文｜工具调用）",
    "contextWindow":1048576,
    "maxTokens":8192,
    "input":["text"],
    "cost":{"input":0,"output":0,"cacheRead":0,"cacheWrite":0},
    "reasoning":true,
    "compat":{
      "thinkingFormat":"deepseek",
      "supportedReasoningEfforts":["low","medium","high","max"],
      "reasoningEffortMap":{"xhigh":"max"}
    }
  }
]' \
--json --merge
# 测试模型是否可用
openclaw infer model run --model sensenova/deepseek-v4-flash --prompt "ping" --local

2.测试模型是否可用

参数	说明
model	固定格式：服务商/模型id
prompt	提示词

openclaw infer model run \
--model sensenova/sensenova-6.7-flash-lite \
--prompt "请回复ping" \
--local

3.修改模型key(交互式)

openclaw models auth paste-api-key --provider sensenova

访问控制台

1. 获取访问地址

# 获取访问地址，一般都是 http://127.0.0.1:18789/
openclaw dashboard --no-open

2. 拼接网关token，获得完整访问链接

token值查询：~/.openclaw/openclaw.json的$.gateway.auth.token

完整地址：http://127.0.0.1:18789?token=8c503c4cf11dc172b72012900677a4f3b460a8c35bcfc503

常用命令

模型

# 查看模型列表
openclaw models list
# 列出sensenova下全部模型
openclaw models list --provider sensenova
# 查询模型服务商配置：openclaw.json 里的
openclaw config get models.providers

# 测试配置模型是否可用（全部模型），Status 显示ok表示可用
openclaw models status --probe
# 测试配置模型是否可用（指定模型），服务商/模型id
openclaw models status --probe sensenova/sensenova-6.7-flash-lite

频道

# 查当前频道
openclaw channels list
# 查询频道绑定agent
openclaw agents bindings
# 飞书频道绑定某个agent
openclaw agents bind --agent main --bind feishu
# 飞书频道解绑某个agent
openclaw agents unbind --agent guzhuang --bind feishu

基础功能

# 修改会话窗口名称
/name 会话名称
# 查询配置文件 openclaw.json 里的 models.providers 配置项
openclaw config get models.providers
# 验证配置文件 openclaw.json 是否正确
openclaw config validate

agent

# 列出所有 agent
openclaw agents list

对接飞书

官方文档

# 飞书配置引导
openclaw channels login --channel feishu
# 获取飞书授权码,首次通过机器人发消息就会有授权码
openclaw pairing approve feishu VLL43MY2
# 重启网关
openclaw gateway restart