5-1、配置不同的【飞书群】对应不同的Agent

单个 Agent 处理多类任务会带来非常严重的记忆负担，由此会极大程度拉高 Token 消耗。我们要建立的就像一个真正的公司或组织内的成员一样，大家各自处理不同类型的任务，分工明确、职能分化。

1、飞书群和agent规划

2、后台命令和配置方式，编辑/root/.openclaw/openclaw.json

（1）把飞书中的机器人配置到channel中

JSON

"channels": {    
    "feishu": {
      "enabled": true,
      "appId": "cli_a93ead42eb79dbcc",
      "appSecret": "<your-feishu-app-secret-1>",
      "domain": "feishu",
      "groupPolicy": "open",
      "requireMention": false,//在群里不用@虾大人
      "allowFrom": [
        "*"
      ]
    }
  },

（2）在Openclaw中建立agent

Plain

openclaw agents add --workspace /root/.openclaw/workspace-nick nick
openclaw agents add --workspace /root/.openclaw/workspace-flash flash
openclaw gateway restart

（3）在飞书中建群，并把虾大人拉到群里，把群的id配置到bindings中，和agent建立关联

JSON

"bindings": [
    {
      "agentId": "main",
      "match": {
        "channel": "feishu",
        "peer": {
          "kind": "group",
          "id": "oc_cf3fd168f4f5df542a5f34004f3a9f49"
        }
      }
    },
    {
      "agentId": "nick",
      "match": {
        "channel": "feishu",
        "peer": {
          "kind": "group",
          "id": "oc_cf1c0af07f746c04733eca468904a91f"
        }
      }
    },
    {
      "agentId": "flash",
      "match": {
        "channel": "feishu",
        "peer": {
          "kind": "group",
          "id": "oc_ebdb9dd0cfae51988b55208e6f3339f4"
        }
      }
    }
  ],