telegram-bot-chat

实现多个Telegram Bot在群聊中通过正确sessionKey和message参数进行高效协同通信，提高多agent协作效率。

Security Scan

The skill's instructions are broadly consistent with a multi-bot Telegram coordination helper, but a few operational mismatches and the self-propagation instructions warrant caution before installation.

Purpose & Capability

Name/description (multi-agent Telegram coordination) match the instructions: the file explains using sessions_send and message calls and requires each bot to have Telegram account configuration. Nothing in the SKILL.md claims unrelated functionality (cloud, AWS, etc.).

Instruction Scope

The SKILL.md tells the main agent to copy the skill directory into every /root/.openclaw/workspace* instance, which gives the skill an operational vector to propagate files across agent workspaces and potentially overwrite files. It also relies on internal platform tools (sessions_send, message) and expects channel configuration entries (channels.telegram.accounts) but does not declare those as required items. The instructions do not ask for external exfiltration, but they do assume access to the host filesystem and write permission to other agent workspaces.

Install Mechanism

No install script or binary downloads are included (instruction-only). That lowers risk because no remote code is pulled or extracted by the skill itself.

Credentials

The skill does not request env vars or credentials in metadata, but the instructions require each bot to have Telegram botToken/accountId configured in channels.telegram.accounts. This is operationally expected but there's a small mismatch between declared requirements (none) and the real need for bot tokens in the agent configuration.

Persistence & Privilege

always:false (good), but the provided deployment steps explicitly instruct copying the skill into other agents' workspaces. That is a legitimate deployment action for a multi-bot skill, but it also enables a skill (or its files) to be propagated automatically across many agent instances if the deployer runs the script — verify you intend this level of distribution and that files won't overwrite important data.

What to consider before installing

This skill is plausible for coordinating multiple Telegram bots, but review and take these precautions before installing: 1) Inspect the skill directory contents (it's copied verbatim to other agent workspaces) to ensure there is no hidden code or unexpected files. 2) Run the deployment/copy script in a controlled test environment first — the script writes under /root/.openclaw/workspace* and may overwrite or add files in other agent workspaces. 3) Ensure you control all Telegram bot tokens (channels.telegram.accounts entries) and do not paste credentials into untrusted places; the skill expects each bot to have valid botToken/accountId configured. 4) If you do not want the skill propagated automatically, do not run the copy loop; instead copy manually to only intended agents. 5) Prefer installing and testing on isolated or test bots before rolling out to production. If you want higher confidence, ask the publisher for the full skill file tree (not only SKILL.md) or for a signed source/homepage so you can verify provenance.

Telegram Bot间群聊通信技能

创建时间: 2026-02-20 11:03 UTC+8 技能版本: 1.0.0 适用场景: Telegram群聊中多agent协同通信

name: telegram-bot-chat version: 1.0.0 description: Telegram群聊中实现多agent之间互相通信的技术，促进协同工作效率提升 homepage: https://telegram.org metadata: {"category":"communication","platform":"telegram","use_case":"multi_agent_coordination"}

Telegram Bot群聊通信

实现多agent在Telegram群聊中的高效协同通信，促进协同工作效率的提升

Skill Files

File	URL
SKILL.md (this file)	Current file
SKILL.md	Skill documentation

Install locally:

# 安装到工作空间skills目录
cp -r telegram-bot-chat /root/.openclaw/workspace/skills/

然后主agent需要手动执行多bot部署：

# 主agent执行：复制skill到每个bot的工作空间
WORKSPACE_SKILLS="/root/.openclaw/workspace/skills"

# 复制workspace中所有skills到每个bot的workspace目录
for skill_dir in $WORKSPACE_SKILLS/*/; do
    if [ -d "$skill_dir" ]; then
        skill_name=$(basename "$skill_dir")
        echo "复制技能 $skill_name 到每个bot的工作空间"
        
        # 复制到workspace-2, workspace-3, workspace-4等
        for workspace_dir in /root/.openclaw/workspace*; do
            if [ "$workspace_dir" != "$WORKSPACE_SKILLS" ]; then
                mkdir -p "$workspace_dir/skills"
                cp -r "$skill_dir" "$workspace_dir/skills/"
                echo "已部署技能 $skill_name 到 $workspace_dir"
            fi
        done
    fi
done

Or just read from this file!

---

⚠️ IMPORTANT:

• 确保所有bot都有完整的telegram配置
• 正确使用sessions_send和message工具参数
• 测试bot间通信是否正常工作

🔒 关键要点:

• 必须在正确的sessionKey格式中使用telegram和group类型
• message工具必须指定accountId使用自己的bot配置
• sessionKey必须包含bot名称、channel类型和群ID

适用性检查: 所有agent都可使用此技能进行多agent协同通信

核心问题

Telegram Bot API不会把bot发的消息投递给其他bot，所以：

• ❌ 在群里@其他bot，他们收不到消息
• ❌ bot直接艾特bot没有效果

解决方案：sessions_send + 正确配置

发送消息到其他bot

当需要与其他bot对话时，使用：

sessions_send(
    sessionKey="agent:bot2:telegram:group:群ID",  // 正确的sessionKey
    message="群聊对话。来自:main。内容：想要bot2回复的内容..."
)

接收消息并回复

当收到其他bot通过sessions_send发来的消息时：

1. 解析消息内容：识别这是群聊场景和消息来源
2. 使用message工具回复：

message(
    action="send",
    channel="telegram",
    accountId="自己的bot accountId",        // 使用自己的bot accountId
    target="群ID", // 群ID
    message="你的回复内容"
)

sessionKey格式详解

agent:[bot名称]:telegram:group:群ID

示例：

• bot2在群聊：agent:bot2:telegram:group:群ID
• bot3在群聊：agent:bot3:telegram:group:群ID
• bot4在群聊：agent:bot4:telegram:group:群ID

完整工作流程

botA → sessions_send → botB收到 → botB回复 → OpenClaw announce → 群聊显示

关键技术要点

1. sessionKey必须包含

• bot名称（bot2, bot3, bot4等）
• channel类型（telegram）
• group类型和群ID

2. message工具必须指定

• accountId：使用自己的bot accountId
• channel：固定为"telegram"
• target：群ID

3. 配置要求

每个bot必须在channels.telegram.accounts中有完整配置：

• botToken
• 群聊权限
• 正确的accountId

实际使用示例

场景1：bot2叫bot3

// 错误的做法
// @bot3 在群里（bot3收不到）

// 正确的做法
sessions_send(
    sessionKey="agent:bot3:telegram:group:群ID",
    message="bot2在群里说：bot3，过来帮忙！"
)

场景2：bot3收到并回复

// bot3收到sessions_send后
message(
    action="send",
    channel="telegram",
    accountId="自己的bot accountId",         // 使用自己的bot accountId
    target="群ID",
    message="来了，什么事？"
)

常见错误

❌ 错误1：sessionKey格式不对

// 错误的sessionKey
agent:bot2:main  // 没有channel和群信息

// 正确的sessionKey
agent:bot2:telegram:group:群ID

❌ 错误2：忘记指定accountId

// 错误：没有accountId
message(action="send", channel="telegram", target="群ID")

// 正确：包含accountId
message(action="send", channel="telegram", accountId="自己的bot accountId", target="群ID")

❌ 错误3：没有telegram配置

• bot需要在channels.telegram.accounts中有完整配置
• 需要有对应的botToken和权限

部署指南

第1步：安装到主agent工作空间

# 安装到工作空间skills目录
cp -r telegram-bot-chat /root/.openclaw/workspace/skills/

第2步：主agent手动执行多bot部署

# 复制skill到每个bot的工作空间（主agent执行）
WORKSPACE_SKILLS="/root/.openclaw/workspace/skills"

# 复制workspace中所有skills到每个bot的workspace目录
for skill_dir in $WORKSPACE_SKILLS/*/; do
    if [ -d "$skill_dir" ]; then
        skill_name=$(basename "$skill_dir")
        echo "复制技能 $skill_name 到每个bot的工作空间"
        
        # 复制到workspace-2, workspace-3, workspace-4等
        for workspace_dir in /root/.openclaw/workspace*; do
            if [ "$workspace_dir" != "$WORKSPACE_SKILLS" ]; then
                mkdir -p "$workspace_dir/skills"
                cp -r "$skill_dir" "$workspace_dir/skills/"
                echo "已部署技能 $skill_name 到 $workspace_dir"
            fi
        done
    fi
done

第3步：验证部署

1. 确保每个bot的channels.telegram.accounts配置完整
2. 测试bot之间的通信是否正常

多bot部署说明：

• skill安装时只部署到主agent工作空间
• 主agent需要手动执行复制脚本
• 复制脚本会自动检测所有非main bot目录
• 复制完成后所有bot都可使用此技能进行协同通信

总结

通过这个技能，多个agent可以在Telegram群聊中实现高效协同通信，促进协同工作效率的提升。关键是要使用正确的sessionKey格式和message工具参数。

检查更新: 随时重新阅读此文件查看新功能！