基础环境中小企业建议本地化部署推荐mac mini可以imessage推荐sudo用户非root进行安装brew需要sudo用户权限不允许root用户安装x86推荐rocky9.6系统软件包server with guiclawhub官方skill市场认证需要github.com的帐号需要科学上网本地模型算力推荐单独用1个节点装我用的是24G显存的显卡尽量不要防到openclaw所在硬件ollama有时候会全部跑cpu影响openclaw的响应速度openclaw最低配置4G内存2核17G硬盘/boot/efi 2G/ 15G以上.openclaw安装切换国内npm源npm install -g openclawlatest --registry https://registry.npmmirror.com ##阿里云没啥说的官网一键安装curl -fsSL https://openclaw.ai/install.sh | bash也可以curl https://openclaw.ai/install.sh -o ./oepnclaw-install.sh到本地后chmod 755 ./oepnclaw-install.sh再安装网络、代理及设备approvesystemctl --user enable openclaw-gatewaysystemctl --user start openclaw-gatewayopenclaw config set gateway.mode localopenclaw config set gateway.bind lanopenclaw config set gateway.auth.token ****bfdb945c72dd7a81****4aa74a6af4def87f7601****openclaw config set gateway.controlUi.allowedOrigins [http://localhost:18789, http://127.0.0.1:18789, http://192.168.254.196:18789, https://192.168.254.196] --jsonopenclaw config set gateway.trustedProxies [127.0.0.1, ::1]配置nginx反向代理openclaw gateway stopopenclaw gateway run操作电脑端访问https://192.168.254.196/#tokens****bfdb945c72dd7a81****4aa74a6af4def87f7601****openclaw devices listopenclaw devices approve ***agent(模型api接入openclaw onboard然后配置modelollama配置范例ollama侧注意配置OLLAMA_CONTEXT_LENGTH64000上下文长度受模型能力、内存、显存影响尽可能大OLLAMA_KEEP_ALIVE-1让需要run的模型一直活动不会自动关闭openclaw节点设置环境变量export OLLAMA_API_KEYollama建议直接修改配置文件通过openclaw onboard配置的无法使用~/.openclaw/openclaw.json中增加models: {providers: {ollama: {baseUrl: http://192.168.254.198:11434,apiKey: ollama,api: ollama,models: [{id: qwen3.5:9b,name: qwen3.5:9b,reasoning: true,input: [text,image],cost: {input: 0,output: 0,cacheRead: 0,cacheWrite: 0},contextWindow: 262144}]}}},agents: {defaults: {model: {primary: ollama/qwen3.5:9b},workspace: /home/claw/.openclaw/workspace,compaction: {mode: safeguard},maxConcurrent: 4,subagents: {maxConcurrent: 8},sandbox: {mode: off}}},~/.openclaw/agents/main/agent/models.json中增加providers: {ollama: {baseUrl: http://192.168.254.198:11434,apiKey: ollama,api: ollama,models: [{id: qwen3.5:9b,name: qwen3.5:9b,reasoning: true,input: [text,image],cost: {input: 0,output: 0,cacheRead: 0,cacheWrite: 0},contextWindow: 262144,maxTokens: 8192,api: ollama}]}}报HTTP 404: 404 page not found错误检查 Ollama 是否运行在终端输入 ollama list如果能列出模型说明 Ollama 没问题。检查 baseUrlopenclaw.json 里的 baseUrl 是不是写了 http://192.168.254.198:11434 ollama千万不要加 /v1。检查 api 类型是不是 api: ollama不是 openai-responses。channel通信客户端接入及初始化telegram参考文档https://cloud.tencent.com/developer/article/2626214app端联系BotFather通过/newbot安装提示创建机器人openclaw配置telegram的channel并绑定HTTP API的keyapp端联系机器人发任意消息处罚验证码openclaw配置approve:openclaw pairing approve telegram 6H****4Lapp端联系机器人发任意消息触发初始化配置IDENTITY.md和USER.md任一channel第一次配置成功才需要后续消息、后续channel配置则不会处罚初始化开始折腾吧飞书参考文档https://www.feishu.cn/content/article/76025192394459742052个重要网站飞书开放平台https://open.feishu.cn/app应用开发和配置删除应用需要提前到管理后台关停应用飞书管理后台https://ycnkgdgecckj.feishu.cn/admin/index应用管理——企业内成员启停应用。先操作开放平台详见参考文档然后操作openclaw配置如下openclaw config set agents.defaults.sandbox.mode off #非容器环境关闭sandboxopenclaw config set channels.feishu.appId cli_****2e2904f8****openclaw config set channels.feishu.appSecret ****aZEKoOddnoM3Fe3sHboKVRoC****openclaw config set channels.feishu.enabled trueopenclaw config set channels.feishu.connectionMode websocketopenclaw config set channels.feishu.dmPolicy pairingopenclaw config set channels.feishu.groupPolicy allowlistopenclaw config set channels.feishu.requireMention trueopenclaw config set channels.feishu.tools.doc disable在飞书app上选中机器人发消息或/create创建验证码有效期5-10分钟然后在openclaw上approve验证码在飞书app上聊天触发初始化配置IDENTITY.md和USER.mdskill基础配置及安装npm install clawhub ##重新登录后命令才生效npx skills ##执行后需要输入y确认后继续clawhub login ##登录到clawhub可以使用github.com的帐号密码。pip install uv ##需要提前安装python3-pip/bin/bash -c $(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh) ##安装brewclawhub install 技能名 开始折腾吧