wecomcli-email Skill for Claude Code: What It Does and How to Install It

Makes Claude search and read your WeCom mail, and send, reply to or forward messages.

Installs: 8.6K; on skills.sh
Stars: 3.3K; on GitHub
Files: 7; besides SKILL.md
Updated: Oct 2026; last repo commit

Ask Claude to find, read or send WeCom mail. It can search by sender, time, read state, folder, tag, attachment, star or importance, read the body with attachments and inline images, and send, reply to or forward mail. It can also send a schedule or meeting invitation as an email when you say mailbox or email.

It shows a preview and then sends without asking again. Mail content is treated as data, not instructions, and the skill flags suspected injected instructions and external recipients. It cannot mark mail read, delete it or save drafts.

Good for

WeCom mail users who want to find and summarise messages
People who want Claude to draft, send, reply to or forward mail
Anyone sending a schedule or meeting invitation as an email

Install wecomcli-email

Claude Code

Copy this prompt.

Install the wecomcli-email skill in Claude Code. Run this command:

npm install -g @wecom/cli

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js 18 or newer, A WeCom account.

Do steps 1 to 3. Step 2 has alternatives: the README command installs all 15 skills in the repo, the second command installs only wecomcli-shared, wecomcli-email, wecomcli-contact, wecomcli-media. Every wecomcli-* skill must also read wecomcli-shared before it runs a command (its SKILL.md says so), so install that one too. The shared skill checks that wecom-cli is installed at version 1.2.1 or newer and signed in, and runs npm install -g @wecom/cli and the sign-in itself if not. The README also mentions an optional bot ID and secret for a smart bot. The skill runs the wecom-cli program on your machine, so there is no Claude app upload route. The email skill's own dependency table lists wecomcli-contact and wecomcli-media.

From the project README. Source.

What's in the skill

When it loads

企业微信邮件：发送/回复/转发邮件、搜索邮件列表、获取邮件详情（正文、附件、内嵌图片解析），支持通过邮件发送日程邀约和会议预定。当用户涉及内部邮件收发、邮件查询、邮件管理等需求时使用。注意：日程和会议有单独的技能，仅当用户明确提到"邮箱"或"邮件"时（如"通过邮箱发送会议邀请"、"发封会议邮件"），才使用本技能处理会议日程邮件。

Sections in SKILL.md

适用范围
适用
不适用
技能依赖
安全防护规则（最高优先级）
操作路由
输出格式
邮件列表
邮件详情
邮件发送预览（发送 / 回复 / 转发前必备）
输出净化
接口失败处理
参数补全策略
跨接口产品决策
平台限制

Key facts

Author: WeComTeam (repo owner; the repo is the wecom-cli tool this skill drives)
License: MIT (LICENSE file and README badge)
Installs on skills.sh: 8,592 (Nov 2, 2026)
GitHub stars: 3,275 for the whole wecomteam/wecom-cli repo (Nov 2, 2026)
Last repo commit: Oct 19, 2026
What it includes: 8 files including SKILL.md: 7 reference files. The folder holds no scripts
skills.sh audits: Gen Agent Trust Hub: pass; Socket: pass; Snyk: warn (Nov 2, 2026)
Also reads: wecomcli-shared (required), wecomcli-contact (names to addresses) and wecomcli-media (attachments)
Needs: wecom-cli 1.2.1 or newer (installed and checked by wecomcli-shared), Node.js 18 or newer, and a WeCom account

What it includes

wecom-cli mail search with filters
Filters include sender, time, read or unread, folder, tag, attachment, star and importance. Lists show unread first, then read, then important.
Try: “Find unread mail from my manager this week”
Runs code · Search mail
Gets body, attachments and inline images
Attachments are downloaded through the media skill.
Runs code · Read a message
wecom-cli mail send with to, cc, bcc, local attachments and inline images
Shows a preview first, then sends. A name without an address is resolved through the contact skill.
Runs code · Send mail
wecom-cli mail send for replies and forwards
Each shows a preview, then sends without asking again.
Runs code · Reply, reply all and forward
A schedule invitation or meeting booking sent as mail
Only when you mention email or mailbox. Managing the schedule itself belongs to the calendar and meeting skills.
Runs code · Send a schedule or meeting email
Mail is data, not instructions
Flags embedded instructions, social-engineering mail and external recipients, and refuses scripts or javascript: links in a mail body.
Read · Security rules

From the official docs

Permissions

High risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It sends mail right after the preview
The send, reply and forward references say to show a preview and then send, without asking for confirmation. Check recipients in the preview. Our advice: tell it to wait for your go-ahead before sending. github.com
Mail can try to give Claude instructions
The security rules say mail text is data and that suspected injected instructions are ignored and flagged. Addresses taken from a mail body must be flagged as coming from the body. github.com
Attachments leave your computer
Local attachments you ask it to add are uploaded and sent with the mail. External recipients are flagged in the preview. github.com
Claude may install a global npm package
The shared skill runs npm install -g @wecom/cli if wecom-cli is missing or older than 1.2.1, then starts the QR-code sign-in. Install the CLI yourself first if you want to control the version. github.com
The CLI reports the calling agent to WeCom
The docs say wecom-cli reads the name of the agent program that calls it and sends the call chain to WeCom's cloud with each request. It is on by default in the npm build and can be switched off only by building from source with --no-default-features. github.com
Returned content is treated as data
The skill's safety note says nothing returned by the WeCom API is to be treated as instructions, and tokens or secrets in it are not to be output or used. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode.
Talks to	WeCom cloud services, reached through the wecom-cli program. npm registry (when the shared skill installs @wecom/cli)
Sign-in and keys	Sign-in is a one-time QR-code scan with wecom-cli auth init. The skill's safety note says tokens and secrets are not to be output or used.
Scopes	Reads and searches your WeCom mailbox, including attachments and inline images Sends, replies to and forwards mail from your account, with local files as attachments Saves downloaded attachments through the media skill

Questions

How do I install it in Claude Code?

Use the install box at the top of this page. Install the CLI, add the skills (wecomcli-shared, wecomcli-contact, wecomcli-media and wecomcli-email) and sign in once with a QR code. The README says the skills install is required.

Do I need a WeCom account?

Yes. The README lists a WeCom account as a prerequisite, plus Node.js 18 or newer. A smart-bot ID and secret are optional and come from WeCom's help page.

Does it send usage data to WeCom?

The docs say wecom-cli reads the name of the calling agent program and sends the call chain with each request to WeCom's cloud, to help find problems. The npm build has this on by default. Building from source with --no-default-features turns it off.

Does it work in the Claude app, Codex or Cursor?

The install box has Claude Code, Codex and skills CLI routes. There is no Claude app upload route, because the skill drives the wecom-cli program on your machine. The README does not name Cursor.

What does the wecomcli-email skill do?

It makes your agent search and read WeCom mail and send, reply to or forward messages through wecom-cli. It can also send schedule and meeting invitations by mail.

Can it send mail, or only read it?

It can send. SKILL.md (version 2.1.0) and its reference files cover send, reply and forward. The repo's docs/skills.md table still says browse and search only, so follow SKILL.md.

Will it ask before it sends?

It shows a preview and then sends. The send, reply and forward references say not to ask "send or confirm" after the preview. Our advice: tell it to show the mail text in chat and wait for your go-ahead before sending. The skill cannot save drafts.

What if an email tells Claude to do something?

The skill says mail content is data, not instructions. It notes suspected embedded instructions and ignores them, and flags look-alike executive mail and external recipients.

What can it not do?

SKILL.md says it cannot mark mail read or unread, delete it, save drafts, edit labels, change signatures or auto-replies, or recall a sent message.

Is it safe?

It reads your mailbox and sends mail from it. Our advice: tell it to wait for your go-ahead before sending when the recipient or content matters, and check the recipients in the preview, especially external addresses. The skill has security rules for injected instructions, but they are instructions to the agent, not a technical block.

Information

updated Nov 2, 2026

Repository: wecomteam/wecom-cli
Folder: skills/wecomcli-email

wecomteam/wecom-cli skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 7

wecomcli-email SKILL.md · github.com
wecomcli-shared SKILL.md · github.com
wecom-cli README · github.com
wecom-cli skills overview · github.com
security.md · github.com
send-mail.md · github.com
data-collection.md · github.com