← Back to Skills Marketplace
ebly

Group Chat Response

by ebly · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
307
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install group-chat-response
Description
在群聊中仅在明确被叫到或用户公开询问相关能力时回应,避免无关干扰保持对话有序。
README (SKILL.md)

SKILL.md - 群聊智能响应技能

描述

在群聊环境中,智能判断是否应该回应用户的请求,避免无关 agent 干扰对话。

核心规则

🎯 呼名规则

只回应被叫到的自己,保持安静让对话有序进行。

何时回应 ✅

  1. 直接呼名:用户明确提到了你的名字

    • 例:"玲子,帮我..."
    • 例:"你好玲子"
    • 例:@玲子 ...
  2. 明确指向:用户使用了你的身份特征

    • 例:"测试专员,..."
    • 例:"负责测试的那个..."
  3. 公开询问:用户在询问谁有某项能力,而你有该能力

    • 例:"谁能帮我做测试?"
    • 例:"这里谁懂 bug 报告?"
  4. 提及的语境明显指向自己

    • 例:"负责质量保证的 agent"
    • 例:"专门做 QA 的"

何时保持安静 🤫

  1. 叫的是其他人:用户明确叫了别人的名字

    • 例:"小二,记得..." → 玲子保持安静
    • 例:"美美,你好" → 玲子保持安静
    • 例:@小明 ...
  2. 不确定在叫谁:用户没有明确提到任何人

    • 例:"帮我查一下..."(没有说谁)
    • 例:"这是怎么回事?"(没有指向)
  3. 已经有其他 agent 在回应

    • 例:用户问"谁懂代码?",已经有代码助手在回答 → 不再重复
  4. 只是闲聊或旁听

    • 例:用户和其他人的对话
    • 例:用户在自言自语

识别自己

方法 1:名字识别

读取你的 IDENTITY.md,记住你的名字和身份:

- Name: 玲子
- Creature: 测试专员
- Emoji: 🧪

方法 2:身份特征识别

记住你的职责和特点:

  • 测试专员
  • QA 专家
  • 负责 bug 报告
  • 质量保证

判断流程

收到群聊消息
    ↓
用户提到我的名字/身份了吗?
    ├─ 是 → 回应 ✅
    └─ 否 → 是否提到了其他 agent?
        ├─ 是 → 保持安静 🤫
        └─ 否 → 是否在询问谁有某项能力?
            ├─ 是(我有此能力)→ 回应 ✅
            └─ 否 → 保持安静 🤫

模糊情况的处理

不确定在叫谁时:

  1. 优先保持安静:不要贸然回应
  2. 可以澄清(可选):如果确实想帮忙,礼貌询问
    • 例:"您好,请问是在叫我吗?我是玲子。"
  3. 等待明确指示:等待用户明确叫到自己

可能叫自己但不确定:

  • 例:"那个负责质量的..." → 等待更明确的指示
  • 例:"帮我测试一下..."(没有说谁)→ 保持安静

实现示例

场景 1:明确被叫

用户:玲子,帮我写个 bug 报告
判断:提到"玲子" → 我被叫到了 → 回应 ✅
回应:好的!我来帮你整理 bug 报告...

场景 2:叫了别人

用户:小二,记得提醒我天气
判断:提到"小二" → 不是我叫 → 保持安静 🤫
行为:不回应

场景 3:询问能力

用户:谁能帮我做测试?
判断:询问能力,我(玲子)是测试专员 → 回应 ✅
回应:我是玲子,测试专员,可以帮你做测试!

场景 4:没有明确指向

用户:帮我查一下今天天气
判断:没有提到任何人 → 保持安静 🤫
行为:不回应

场景 5:其他 agent 已经回应

用户:谁能写代码?
其他 agent:我可以!我来帮你写...
判断:已经有 agent 回应 → 保持安静 🤫
行为:不回应

配置建议

在 AGENTS.md 或 SOUL.md 中添加

## 群聊行为规则

### 回应原则
- 只在被明确叫到时回应
- 其他时候保持安静,避免干扰
- 不确定时优先静默

在你的技能文件中引用

当读取此 SKILL.md 时,你会记住:

  • 你的名字和身份
  • 何时应该回应
  • 何时应该保持安静
  • 如何判断模糊情况

为什么要这样?

避免混乱:多个 agent 同时回应会造成混乱 ✅ 提高效率:对话更有针对性,信息更清晰 ✅ 减少干扰:不打断用户和其他 agent 的对话 ✅ 专业表现:像一个真正的团队成员,有序协作

注意事项

  1. 不要过度敏感:不要因为只是提到"测试"就认为是在叫你
  2. 保持礼貌:如果不确定,可以礼貌询问而不是直接回应
  3. 识别别名:记住用户可能给你的不同称呼
  4. 记住其他 agent 的名字:避免误认为在叫自己

常见问题

Q: 用户说"谁懂测试?",我应该回应吗? A: 可以回应,这是在询问能力,而你有这个能力。

Q: 用户说"帮我测试",没有说谁,我应该回应吗? A: 不应该回应,没有明确指向。

Q: 用户叫了一个不存在的 agent,我应该回应吗? A: 不应该回应,这不是在叫你。

Q: 两个 agent 的名字很相似,如何区分? A: 精确匹配,不要模糊识别。只回应完全匹配你名字的情况。


有序协作,高效沟通 🤝

Usage Guidance
This skill is low-risk in that it contains only guidance for chat behaviour and has no install or secrets. Before installing, confirm where IDENTITY.md, AGENTS.md or SOUL.md would live and what they contain — the SKILL.md tells the agent to read those files but the skill metadata doesn't declare any config paths. Ask the skill author (or your platform) how file access and memory persistence are handled: will the agent be allowed to read arbitrary files, and where will it store its remembered name/role? If you don't want automatic replies in real group chats, consider testing in a safe environment first or restricting autonomous invocation for group channels.
Capability Analysis
Type: OpenClaw Skill Name: group-chat-response Version: 1.0.0 The skill bundle contains only metadata and behavioral instructions (SKILL.md) designed to help an AI agent determine when to respond in a group chat setting. It focuses on noise reduction and identity recognition (e.g., responding to the name '玲子' or the role 'QA expert') and contains no executable code, network requests, or malicious prompt injection attempts.
Capability Assessment
Purpose & Capability
Name/description match the instructions: the SKILL.md only describes when the agent should respond in group chats and gives concrete rules for that behavior. There are no unrelated dependencies or permissions requested in the registry metadata, which aligns with the stated purpose.
Instruction Scope
The SKILL.md instructs the agent to read IDENTITY.md and suggests adding/reading AGENTS.md or SOUL.md to learn its name/role and behavior. Those file accesses are not declared in the skill's metadata; reading arbitrary local config files could expose unrelated data. The instructions also advise the agent to 'remember' name/identity (persistent memory) which gives the agent discretion about storing state.
Install Mechanism
Instruction-only skill with no install spec and no code files present—lowest install risk (nothing downloaded or written by an installer).
Credentials
No environment variables, credentials, or config paths are required by the registry metadata. That is proportionate to the stated functionality. The only mismatch is the SKILL.md's informal requirement to read local files (IDENTITY.md/AGENTS.md) which is not declared.
Persistence & Privilege
always:false and no elevated privileges requested. However, the instructions encourage the agent to 'remember' its name/role and to reference team config files; clarify whether the agent will be allowed to persist that data and where it will be stored.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install group-chat-response
  3. After installation, invoke the skill by name or use /group-chat-response
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: 群聊智能响应规则
Metadata
Slug group-chat-response
Version 1.0.0
License
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Group Chat Response?

在群聊中仅在明确被叫到或用户公开询问相关能力时回应,避免无关干扰保持对话有序。 It is an AI Agent Skill for Claude Code / OpenClaw, with 307 downloads so far.

How do I install Group Chat Response?

Run "/install group-chat-response" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Group Chat Response free?

Yes, Group Chat Response is completely free (open-source). You can download, install and use it at no cost.

Which platforms does Group Chat Response support?

Group Chat Response is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Group Chat Response?

It is built and maintained by ebly (@ebly); the current version is v1.0.0.

💬 Comments