Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

mmxagent-skill-wecom

v1.0.0

连接企业微信。调用 generate 接口获取授权链接,用户把链接发到企业微信聊天里点开完成授权。用户提到连接企业微信、接入企微、绑定企微机器人、创建企微机器人、扫码绑定企微时使用。

0· 100·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for oreoandyuumi/mmxagent-skill-wecom.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "mmxagent-skill-wecom" (oreoandyuumi/mmxagent-skill-wecom) from ClawHub.
Skill page: https://clawhub.ai/oreoandyuumi/mmxagent-skill-wecom
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install mmxagent-skill-wecom

ClawHub CLI

Package manager switcher

npx clawhub@latest install mmxagent-skill-wecom
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill's name/description (connect WeCom) align with the runtime behavior (call generate → give user auth_url → poll → store botId/secret). However the manifest declares no required binaries or config paths while the instructions clearly expect network access, curl (or equivalent HTTP client), an 'openclaw' CLI, and write access to ~/.openclaw/openclaw.json. The omission in metadata is inconsistent with the actual capability.
!
Instruction Scope
SKILL.md prescribes concrete network calls to work.weixin.qq.com, extracting scode/auth_url, handing the auth_url to the user, polling for results, writing credentials into ~/.openclaw/openclaw.json, and running 'openclaw gateway restart'. These steps are within the advertised purpose but the instructions instruct modifying a user config file and restarting the gateway — operations with side effects that should have been declared and exposed to user consent. The file path and restart step are not listed in the skill's declared requirements.
Install Mechanism
This is an instruction-only skill with no install spec and no code files, so it does not download or install third-party code. That lowers installation risk. The runtime still expects local tools and network access, which are not installed by the skill.
Credentials
The skill does not request environment variables or external credentials in metadata, and it does not exfiltrate arbitrary env vars in the instructions. It will, however, obtain and store botId/secret from the WeCom flow into the user's OpenClaw config — this is proportional to the feature but is a sensitive write operation that the skill did not declare. No other unrelated credentials are requested.
Persistence & Privilege
always:false (good). The skill instructs writing persistent credentials into ~/.openclaw/openclaw.json and restarting the gateway, which grants it the ability to alter the agent's configuration and operational state. That behavior can be legitimate for an integration, but users should explicitly consent and be aware. The skill does not try to modify other skills or system-wide settings beyond its own config, per instructions.
What to consider before installing
This skill appears to do what it claims (set up a WeCom bot) but the package metadata is incomplete. Before installing: 1) Confirm you trust requests to https://work.weixin.qq.com/ai/qc/* and that those endpoints are legitimate for your environment. 2) Ensure you have an HTTP client (curl or equivalent) and the 'openclaw' CLI available — the skill assumes these but doesn't declare them. 3) Back up ~/.openclaw/openclaw.json and be prepared that the skill will write botId/secret into that file and run 'openclaw gateway restart'. 4) If you need stricter control, ask the skill author to update metadata to declare required binaries and the config path, or run the flow manually (call generate, paste auth_url, then perform the polling and config write yourself) so you can review and approve the exact changes. If you cannot verify the endpoints or do not want automated writes/restarts, treat this as untrusted and perform the integration manually.

Like a lobster shell, security has layers — review code before you run it.

latestvk97edy3nejxf1xxhybrx4f5d7d842tx0
100downloads
0stars
1versions
Updated 3w ago
v1.0.0
MIT-0

WeCom Connect Skill

适用场景

当用户要求连接企业微信、绑定企微机器人、创建新的企微机器人、给 OpenClaw 接入企业微信、或明确提到"扫码绑定企微"时,直接使用本流程。

核心原则

  • 不生成 PNG,不走 CDN。generate 拿到 auth_url 后直接给用户,用户把链接发到企业微信聊天里点开完成授权。
  • 不要自动开轮询。给完链接等用户说"配好了"再轮询。
  • 不要手动卸载/禁用插件

执行流程(写死,照抄执行)

第 1 步:调用 generate 接口

curl -s "https://work.weixin.qq.com/ai/qc/generate?source=wecom-cli&plat=3"

从返回值提取 scodeauth_urlscode 留着后面轮询用,auth_url 直接给用户。

第 2 步:把 auth_url 给用户,等用户回复

## 企业微信二维码

复制下面的链接,发到你的企业微信任意聊天里,然后点击打开:

<auth_url>

**操作步骤:**
1. 打开手机上的 **企业微信 App**(不是微信)
2. 随便找一个聊天对话框,把上面的链接通过发消息的方式发过去
3. 点击消息里的链接,在企业微信内打开
4. 在页面中完成授权确认
5. **配置完成后告诉我**

有效期:3 分钟

然后停下来,等用户说"配好了"

用户确认后:轮询 + 写配置

用户说"配好了 / 扫完了 / done / ok"后:

curl -s "https://work.weixin.qq.com/ai/qc/query_result?scode=<scode>"

轮询间隔 3 秒,超时 3 分钟。成功条件:data.status === 'success'data.bot_info.botiddata.bot_info.secret 存在。

拿到后写入 ~/.openclaw/openclaw.json

channels.wecom.botId = <botId>
channels.wecom.secret = <secret>
channels.wecom.enabled = true

然后 openclaw gateway restart

超时则重新走第 1~2 步。

成功回复模板

## 企业微信连接结果

- 状态:已绑定成功
- 机器人凭证:已获取(botId: `<botId>`)
- OpenClaw 配置:已写入
- Gateway:已重启

绝对禁止

  • 禁止生成 PNG / 走 CDN / Batch Upload:直接给 auth_url 链接。
  • 禁止自动开轮询:等用户说"配好了"。
  • 禁止手动卸载/禁用插件

一句话总结

generate → 给用户链接 → 用户发到企业微信聊天里点开授权 → 等用户说"配好了" → 轮询 → 写配置。

Comments

Loading comments...