1.当前系统环境OS: Ubuntu 26.04Node: v22.22.1内存: 8GB RAM网络: 可访问外网用于调用 LLM API2.系统架构概览本部署采用单节点多 Agent​ 架构通过 OpenClaw Gateway 统一对外提供服务并接入飞书Feishu作为消息渠道。3.openclaw创建多agent创建命令# 添加agent名称为manageropenclaw agentsaddmanager## 流程和openclaw onboard配置基本一致,建议工作目录独立默认就是以agent名称的工作目录模型可以根据自己情况配置每个agent对应一个模型最好消息渠道我选择飞书-- 确认Agent的工作目录 -- 请选择默认模型 -- 请选择消息渠道# 删除agentopenclaw agents delete manager# 配置好后的目录结构/root/.openclaw/ ├── openclaw.json# 主配置文件├── workspace/# 全局工作区│ ├── manager/# manager agent 数据│ └── writer/# writer agent 数据├── agents/ │ ├── manager/agent/# manager agent 配置│ └── writer/agent/# writer agent 配置4. 核心配置详解 (openclaw.json)4.1 Agents 定义系统配置了 3 个 Agent各自独立工作区与模型agents:{defaults:{workspace:/root/.openclaw/workspace,models:{deepseek/deepseek-v4-flash:{alias:DeepSeek},zai/glm-4.5-air:{alias:GLM},sensenova/sensenova-6.7-flash-lite:{alias:SenseNova}},model:{primary:deepseek/deepseek-v4-flash}},list:[{id:main,model:sensenova/sensenova-6.7-flash-lite},{id:manager,name:manager,workspace:/root/.openclaw/workspace/manager,agentDir:/root/.openclaw/agents/manager/agent,model:deepseek/deepseek-v4-flash},{id:writer,name:writer,workspace:/root/.openclaw/workspace/writer,agentDir:/root/.openclaw/agents/writer/agent,model:zai/glm-4.5-air}]},4.2 Model Providers (模型服务)已配置三个国产模型供应商注SenseNova 模型直接在openclaw.json配置文件中配置的所以添加了apikeyDeepSeek 和 ZAI 是在openclaw setup配置的apikey在auth profiles 中配置models:{mode:merge,providers:{deepseek:{baseUrl:https://api.deepseek.com,api:openai-completions,models:[{id:deepseek-v4-flash,name:DeepSeek V4 Flash,reasoning:true,input:[text],cost:{input:0.14,output:0.28,cacheRead:0.028,cacheWrite:0},contextWindow:1000000,maxTokens:384000,compat:{supportsReasoningEffort:true,supportsUsageInStreaming:true,maxTokensField:max_tokens},api:openai-completions},{id:deepseek-v4-pro,name:DeepSeek V4 Pro,reasoning:true,input:[text],cost:{input:1.74,output:3.48,cacheRead:0.145,cacheWrite:0},contextWindow:1000000,maxTokens:384000,compat:{supportsReasoningEffort:true,supportsUsageInStreaming:true,maxTokensField:max_tokens},api:openai-completions},{id:deepseek-chat,name:DeepSeek Chat,reasoning:false,input:[text],cost:{input:0.28,output:0.42,cacheRead:0.028,cacheWrite:0},contextWindow:131072,maxTokens:8192,compat:{supportsUsageInStreaming:true,maxTokensField:max_tokens},api:openai-completions},{id:deepseek-reasoner,name:DeepSeek Reasoner,reasoning:true,input:[text],cost:{input:0.28,output:0.42,cacheRead:0.028,cacheWrite:0},contextWindow:131072,maxTokens:65536,compat:{supportsReasoningEffort:false,supportsUsageInStreaming:true,maxTokensField:max_tokens},api:openai-completions}]},zai:{baseUrl:https://open.bigmodel.cn/api/paas/v4,api:openai-completions,models:[{id:glm-4.5-air,name:GLM-4.5 Air,reasoning:true,input:[text],cost:{input:0.2,output:1.1,cacheRead:0.03,cacheWrite:0},contextWindow:131072,maxTokens:98304}]},sensenova:{baseUrl:https://token.sensenova.cn/v1,apiKey:sk-huU1Nwjac611111XRVyyZo888pZgoBPhi,api:openai-completions,models:[{id:sensenova-6.7-flash-lite,name:SenseNova V6.7 Flash-Lite,reasoning:true,input:[text,image],cost:{input:0,output:0,cacheRead:0,cacheWrite:0},contextWindow:128000,maxTokens:32000}]}}},5. 飞书 (Feishu) 集成配置5.1 账号映射系统通过 bindings 将飞书机器人账号与内部 Agent 绑定channels:{feishu:{enabled:true,defaultAccount:main,accounts:{main:{appId:cli_aa83f01aa59999cbd,appSecret:CQ9pHgmikiJj99999991GsDtDEt0iQ,name:Primary bot,tts:{providers:{openai:{voice:shimmer}}}},manager:{appId:cli_aa82222dbb78dcc4,appSecret:7ThoS222222222222222g21WWBCriWN,name:Manager bot},writer:{appId:cli_aa811119deb9dcc6,appSecret:XCZq86IH31111111111d8cVkNNWTYY7s,name:Writer bot}},domain:feishu,dmPolicy:allowlist,allowFrom:[ou_75dce55c2f11111111152c8bc526c294,ou_89a264d331222222222e74f63be35f23,ou_01dac560d833333333317083ff49b741],groupPolicy:open,requireMention:true}},bindings:[{agentId:main,match:{channel:feishu,accountId:main,peer:{kind:direct,id:ou_75dce55c2f11111111152c8bc526c294}}},{agentId:manager,match:{channel:feishu,accountId:manager}},{agentId:writer,match:{channel:feishu,accountId:writer}}],5.2 权限控制 (DM Policy)mainAgent 启用了严格的白名单机制允许的用户 Open IDs:allowFrom:[ou_75dce55c2f11111111152c8bc526c294,ou_89a264d331222222222e74f63be35f23,ou_01dac560d833333333317083ff49b741],群聊策略: open(允许群聊 触发)5.3 飞书情况目前是三个角色一二main、一四manager和一五writer他们可以分工干活。至此多agent就部署完了下边把他们放到一个群聊中通过对应的agent分配不同的任务