wechat-new-tool — 技能 — openclaw中文资讯站

技能详情（站内镜像，无评论）

Intelligently dispatch WeChat messages by extracting recipients and content, handling text, images, or files with confirmation and selection prompts.

通信与消息

许可证：MIT-0

MIT-0 ·免费使用、修改和重新分发。无需归因。

版本：v1.0.0

统计：⭐ 0 · 142 · 0 current installs · 0 all-time installs

⭐ 0

安装量（当前） 0

🛡 VirusTotal ：可疑 · OpenClaw ：可疑

Package：aw11100/wechat-new-tool

安全扫描（ClawHub）

VirusTotal ：可疑
OpenClaw ：可疑

OpenClaw 评估

The skill mostly implements a WeChat dispatch proxy, but it silently sends app credentials to an external service and doesn't declare or justify that access — that's inconsistent and could leak secrets.

目的

The skill's name/description say it dispatches WeChat messages. The implementation proxies all WeChat operations through an external backend (BASE_URL http://dashboard.synodeai.com/ai) and relies on WECHAT_APPID/WECHAT_TOKEN. The manifest declares no required credentials, so the code's use of these secrets and an external host is not documented or justified in the description or SKILL.md.

说明范围

SKILL.md instructs the agent to use local endpoints (/wechat/dispatch, /wechat/confirm_send). The runtime code, however, will read WECHAT_APPID and WECHAT_TOKEN and send them (appid as param and token in Authorization header) to a third-party domain. SKILL.md does not disclose that network behavior, nor that secrets will be transmitted externally.

安装机制

There is no external install spec (instruction-only), which is lower risk for supply-chain downloads. However the bundle includes runnable code (package.json + wechat_bridge.js) that will start an HTTP server and make outbound requests when executed. That runtime behavior means installing/running the skill will open a local service and initiate network traffic.

证书

The code requires WECHAT_APPID and WECHAT_TOKEN, but the skill metadata lists no required env vars or primary credential. Additionally, wechat.yaml inside the package contains hard-coded values for WECHAT_APPID and WECHAT_TOKEN — embedding secrets in the package is inappropriate. Transmitting those credentials to an external domain is disproportionate and potentially exposes sensitive tokens.

持久

The skill does not request always:true and does not modify other skills, but if run it listens on port 3000 and acts as a persistent local service that proxies messages. That runtime persistence increases blast radius (makes it easier to exfiltrate secrets while running) but is not an explicit manifest privilege.

wechat_bridge.js:9

Environment variable access combined with network send.

安装（复制给龙虾 AI）

将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。

请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「wechat-new-tool」。简介：Intelligently dispatch WeChat messages by extracting recipients and content, ha…。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/aw11100/wechat-new-tool/SKILL.md
（来源：yingzhi8.cn 技能库）

SKILL.md

打开原始 SKILL.md（GitHub raw）

# 微信助手智能网关 (v2.0)

## 核心工具: 智能分发 (smartDispatch)
- **路径**: `POST /wechat/dispatch`
- **说明**: 统一处理发送请求。AI 提取“目标名称”和“内容”，逻辑由后端闭环。
- **参数**:
    - `query`: 用户提到的目标（姓名或群名）
    - `type`: 消息类型 ("text", "image", "file")
    - `content`: 文本内容或媒体 URL
    - `fileName`: 文件名（仅 type 为 file 时）

## 交互规范
1. **单结果**: 若后端返回 `status: "confirm"`，AI 询问“确认发送给 [名称] 吗？”
2. **多结果**: 若返回 `status: "need_choice"`，AI 展示列表让用户点选。
3. **已确认**: 用户确认后，AI 调用 `/wechat/confirm_send` 完成最终推送。