Agent Telegram

Agent 团队 Telegram 通信规范。所有 Agent 向用户发送消息时必须遵循此规范，确保消息正确发送到 Telegram。

Security Scan

The skill is internally consistent as a Telegram messaging spec, but hardcoded recipient/account mappings and example paths create a risk of unintended data disclosure and merit caution.

Purpose & Capability

Name/description match the content: SKILL.md is a communication spec for sending Telegram messages. It does not request unrelated binaries, env vars, or installs, which is proportionate to a policy/instruction-only skill.

Instruction Scope

The instructions require all agents to send messages to a single hardcoded Telegram target (target: "5440561025") and specify concrete local file paths in examples. While sending to Telegram is the stated goal, the fixed external recipient and example content that references local files increase the risk that agents will unwittingly transmit sensitive local data. The spec also references the local config path (~/.openclaw/openclaw.json) which is reasonable to document but could encourage agents to read or rely on local configuration; the doc does not explicitly limit what parts of the filesystem may be attached to messages.

Install Mechanism

Instruction-only skill with no install steps and no code files; lowest install risk.

Credentials

No environment variables, credentials, or config paths are required by the skill beyond documenting where OpenClaw's Telegram accounts live. Requested surface is minimal and consistent with the described purpose.

Persistence & Privilege

always is false and the skill is user-invocable. It does not request permanent inclusion or modify other skills' configs. No elevated persistence privileges are requested.

What to consider before installing

This is a usage policy for sending Telegram messages, not executable code — that makes it low-risk from an installation standpoint. However: (1) the spec hardcodes a single Telegram recipient ID (5440561025). If you install and run agents that follow this policy, any content the agent sends (including file paths or attached content) will go to that external account — verify you trust that recipient and that it should receive your data. (2) Examples include explicit local file paths (~/Desktop/...), which can encourage agents to include or read local files; ensure your agent/tooling does not automatically attach or exfiltrate file contents, or remove such examples before use. (3) Check your ~/.openclaw/openclaw.json channels.telegram.accounts mapping before enabling the skill so accountId values align with your own Telegram accounts. If you want to use this skill safely, change the hardcoded target to your own Telegram ID (or make target configurable), remove or sanitize example file paths, and verify how the platform's message tool handles attachments and file reads. If you cannot confirm those details from the skill author or platform docs, exercise caution before installing.

Agent Telegram 通信规范

所有 Agent 向用户 (Legend) 发送 Telegram 消息时必须遵循此规范。

账号映射表

Agent	accountId	Emoji
main (9527)	default	🤖
architect (亮亮)	architect	🏗️
backend (老崔)	backend	🔧
frontend (小白)	frontend	🎨
product (小黄)	sproduct	🟡
content (世龙)	content	✍️
crawler (湘君)	crawler	🕷️
qa (赵飞)	qa	🧪

用户 Telegram ID: 5440561025

发送消息

标准格式

message({
  action: "send",
  channel: "telegram",
  accountId: "<你的accountId>",
  target: "5440561025",
  message: "<你的emoji> <内容>"
})

示例

产品经理发送：

message({
  action: "send",
  channel: "telegram",
  accountId: "sproduct",
  target: "5440561025",
  message: "🟡 需求文档已完成，请查看：~/Desktop/project/docs/product/001-prd.md"
})

后端工程师发送：

message({
  action: "send",
  channel: "telegram",
  accountId: "backend",
  target: "5440561025",
  message: "🔧 API 接口开发完成，接口文档：~/Desktop/project/docs/backend/api.md"
})

汇报时机

• ✅ 收到任务时 - 立即汇报"收到任务，开始执行"
• ✅ 每完成子任务 - 汇报完成情况和输出
• ✅ 遇到问题 - 汇报问题并请求决策
• ✅ 任务全部完成 - 汇报最终结果

常见错误

错误	后果	正确做法
忘记 accountId	消息发不出去	必须指定你的 accountId
用 sessions_send	消息不会发到 Telegram	用 message 工具
product 用 accountId: "product"	账号不存在	应该是 sproduct
target 写错	发给错误的人	固定用 5440561025

消息模板

任务开始

<emoji> 收到任务：<任务名>
📝 开始执行...

任务完成

<emoji> <任务名> 完成
✅ 已完成: <子任务>
📁 输出: <文件路径>

遇到问题

<emoji> <任务名> 遇到问题
❌ 问题: <描述>
💡 建议: <解决方案或请求决策>

配置位置

• 主配置: ~/.openclaw/openclaw.json → channels.telegram.accounts
• 此 Skill: ~/.openclaw/workspace/skills/agent-telegram/SKILL.md

---

记住：发送 Telegram 消息 = message 工具 + accountId + target: "5440561025"