Hermes Agent连接飞书群聊中不用@无响应问题

最近在使用了一段时间openclaw后,觉得多agent协作非常好用,但有些🔥token,正好最近hermes开源了,就研究了一下,但对飞书的群聊支持需要一些调整,我做了几种情况的案例,供大家参考,希望大家共同推进多agent技术进步,具体调整如下。详细源码见github库。

一、方式1:指定群不用@有响应,其他群需要@才响应

实现在群1中不用@也有反馈,但在其他群聊中仍保持原样,通过@进行返回。经过测试只能通过修改配置文件和feishu.py代码来实现,具体记录如下。

1..env文件修改
复制代码
# 主要增加了 FEISHU_FREE_RESPONSE_CHATS 环境变量
FEISHU_DOMAIN=feishu
FEISHU_CONNECTION_MODE=websocket
FEISHU_ALLOW_ALL_USERS=false
FEISHU_ALLOWED_USERS=群1id,群2id
FEISHU_GROUP_POLICY=open
FEISHU_FREE_RESPONSE_CHATS=群1id # ← 新增:这个群免 @
2.config.yaml文件修改
复制代码
# 其实这两个参数可以去掉,对feishu无效,好像对discord有用,这里可以不做调整
feishu:
  require_mention: false
  free_response_chats: 群1id
3.feishu.py修改

完整代码请访问github库的feishu.py。

复制代码
# 关键调整如下,详见dded by 注释之间内容。
def _should_accept_group_message(self, message: Any, sender_id: Any, chat_id: str = "") -> bool:
        """Require an explicit @mention before group messages enter the agent.
		
		Exception: chats listed in FEISHU_FREE_RESPONSE_CHATS bypass the @mention gate.
		"""
        if not self._allow_group_message(sender_id, chat_id):
            return False
		# added by in 2026-4-22 start=====================
		# Check if this chat is in the free-response allowlist (no @mention required)
        free_response_chats = set(
            item.strip()
            for item in os.getenv("FEISHU_FREE_RESPONSE_CHATS", "").split(",")
            if item.strip()
        )
        if chat_id and chat_id in free_response_chats:
            return True
		# added by in 2026-4-22 end=====================

        # @_all is Feishu's @everyone placeholder --- always route to the bot.
        raw_content = getattr(message, "content", "") or ""
        if "@_all" in raw_content:
            return True
        mentions = getattr(message, "mentions", None) or []
        if mentions:
            return self._message_mentions_bot(mentions)
        normalized = normalize_feishu_message(
            message_type=getattr(message, "message_type", "") or "",
            raw_content=raw_content,
        )
        if normalized.mentioned_ids:
            return self._post_mentions_bot(normalized.mentioned_ids)
        return False
二、方式2:指定群不用@有响应,其他允许群需要@才响应,没配置群一律无响应

实现在群1中不哟用@也有反馈,但在其他允许群聊中仍保持原样,通过@进行返回,但在没有设置允许的其他群没有任何响应(相当于不允许群聊)。经过测试只能通过修改配置文件和feishu.py代码来实现,这种情况修改的内容有些多,具体记录如下。

1..env文件修改
复制代码
# 主要增加了 FEISHU_FREE_RESPONSE_CHATS 环境变量
FEISHU_DOMAIN=feishu
FEISHU_CONNECTION_MODE=websocket
FEISHU_ALLOW_ALL_USERS=false
FEISHU_ALLOWED_USERS=群1id,群2id
FEISHU_GROUP_POLICY=allowlist
FEISHU_HOME_CHANNEL=群1id # 可以不用
2.config.yaml文件修改
复制代码
# 关键修改,增加的群聊规则,支持三种情况
feishu:
  extra:
      # 群聊规则配置
      group_rules:
        群1id:
          policy: open           # 允许所有人
          at_only: false         # ← 免 @ 响应!
        群2id:
          policy: open           # 允许所有人
          at_only: true         # ← 需要 @ 才响应!
      # 默认策略:未配置的群聊全部禁用
      default_group_policy: disabled
3.feishu.py修改

主要修改feishu.py文件,具体如下。完整代码请访问github库的feishu-noreply.py。

复制代码
# 关键调整如下,详见dded by 注释内容。
class FeishuGroupRule:
    """Per-group policy rule for controlling which users may interact with the bot."""

    policy: str  # "open" | "allowlist" | "blacklist" | "admin_only" | "disabled"
	at_only: bool = True  # ← 新增:True=需要@,False=免@响应 added by 
    allowlist: set[str] = field(default_factory=set)
    blacklist: set[str] = field(default_factory=set)

...

def _load_settings(extra: Dict[str, Any]) -> FeishuAdapterSettings:
        # Parse per-group rules from config
        raw_group_rules = extra.get("group_rules", {})
        group_rules: Dict[str, FeishuGroupRule] = {}
        if isinstance(raw_group_rules, dict):
            for chat_id, rule_cfg in raw_group_rules.items():
                if not isinstance(rule_cfg, dict):
                    continue
                group_rules[str(chat_id)] = FeishuGroupRule(
                    policy=str(rule_cfg.get("policy", "open")).strip().lower(),
					at_only=_to_boolean(rule_cfg.get("at_only", True)),  # ← 新增 added by 
                    allowlist=set(str(u).strip() for u in rule_cfg.get("allowlist", []) if str(u).strip()),
                    blacklist=set(str(u).strip() for u in rule_cfg.get("blacklist", []) if str(u).strip()),
                )

def _should_accept_group_message(self, message: Any, sender_id: Any, chat_id: str = "") -> bool:
        """Require an explicit @mention before group messages enter the agent.
		
		Per-group behavior:
        - If group has a rule with policy="disabled": no response at all
        - If group has a rule with at_only=false: no @mention required
        - If group has a rule with at_only=true (default): @mention required
        - If no group rule: fallback to default behavior (@mention required)
		"""
        if not self._allow_group_message(sender_id, chat_id):
            return False
		# added by in 2026-4-22 start=====================
		# Check per-group at_only setting
        rule = self._group_rules.get(chat_id) if chat_id else None
        if rule:
            # at_only=false means respond without @mention
            if not rule.at_only:
                return True
            # at_only=true (default): fall through to @mention check below
		# add by in 2026-4-22 end=====================

        # @_all is Feishu's @everyone placeholder --- always route to the bot.
        raw_content = getattr(message, "content", "") or ""
        if "@_all" in raw_content:
            return True
        mentions = getattr(message, "mentions", None) or []
        if mentions:
            return self._message_mentions_bot(mentions)
        normalized = normalize_feishu_message(
            message_type=getattr(message, "message_type", "") or "",
            raw_content=raw_content,
        )
        if normalized.mentioned_ids:
            return self._post_mentions_bot(normalized.mentioned_ids)
        return False
相关推荐
宝哥聊编程9 天前
主流 Agent 平台对比:Coze、Dify、飞书、扣子、OpenClaw 落地实操手册
前端·人工智能·飞书
guoran_shini10 天前
Agent系统服务化封装和IM接入
飞书
承渊政道13 天前
飞牛NAS部署Hermes AI Agent:接入DeepSeek、微信与公网访问完整教程
人工智能·微信·nas·deepseek·hermes
mudtools13 天前
在飞书 SDK 上用 Microsoft Agent Framework 搭一个能干活的工作搭子
人工智能·ai·.net·飞书·agent
极致了14 天前
飞书插件、RPA、API三条路的真实分界线
飞书·rpa
玩泥巴的15 天前
在飞书 SDK 上用 Microsoft Agent Framework 搭一个能干活的工作搭子
ai·.net·飞书·agent
玩泥巴的16 天前
一份声明,多份产物:飞书 SDK 的 AI 工具面是怎么生成的
ai·.net·飞书·agent
OpenPortal认证计费系统18 天前
基于 飞书认证 通过 华为、H3C 结合 OpenPortal认证计费系统 实现 网络准入 飞书授权 实名认证
华为·飞书·h3c·portal认证·openportal
前进的李工18 天前
飞书企业知识库设计+后端高频八股
数据库·redis·后端·mysql·性能优化·飞书·rag