Mailbox logo

Mailbox

OrganizationPopular
agentscope-ai
mailbox

Use this skill whenever the user needs ANY mailbox/email operation — checking, reading, searching, sending, replying, forwarding, organizing or deleting email, managing threads, connecting a personal mailbox, or registering a new mailbox. This skill is the single entry point for email tasks and orchestrates qwenpawmail-mcp for nine supported personal-mail domains.

Overview

Publisheragentscope-ai
RepositoryQwenPaw
Skill namemailbox
Stars
35.6K
Forks
3.2K
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by agentscope-ai on GitHub. Read the source before you install it.

Installation

Install the Mailbox AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/agentscope-ai/QwenPaw/tree/main/src/qwenpaw/agents/skills/mailbox-en
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/agentscope-ai/QwenPaw.git /tmp/QwenPaw
mkdir -p .claude/skills
cp -r /tmp/QwenPaw/src/qwenpaw/agents/skills/mailbox-en .claude/skills/mailbox
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Mailbox in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Mailbox on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Mailbox is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Mailbox Operations (qwenpawmail-mcp)

Use qwenpawmail-mcp to connect or register a mailbox and perform email operations.

Supported Providers

The managed QwenPaw mailbox workflow supports these nine mail domains:

ProviderDomainsLogin credential
NetEase163.com, 126.com, yeah.net16-character authorization code
Tencentqq.com, foxmail.com16-character authorization code
Sinasina.com, sina.cn16-character authorization code
Alibabaaliyun.comMailbox login password; existing accounts only
Googlegmail.com16-character app password after enabling 2-Step Verification

Enterprise mailboxes, custom domains, and Microsoft mailboxes are not supported by the current managed workflow. Do not try enterprise provider overrides or custom IMAP/SMTP hosts. Ask the user to choose one of the supported personal domains instead.

create_mailbox provides built-in registration guidance only for 163.com, 126.com, yeah.net, qq.com, and foxmail.com. Other supported domains require their official registration flow. New aliyun.com registration is unavailable; only existing accounts can be connected.

Invocation Rule

For any email operation, use this skill as the entry point. Do not invent another email workflow or bypass qwenpawmail-mcp with raw IMAP/SMTP commands.

Credential and Configuration Model

agent.json contains only public mailbox configuration. Its expected mail shape is:

json
{
  "mail": {
    "is_new_account": false,
    "credential": {
      "name": "myaccount",
      "domain": "163.com",
      "provider": ""
    },
    "push": {
      "mode": "off",
      "rules": [],
      "poll_interval_seconds": 120,
      "access_control_enabled": false
    }
  }
}

The sensitive fields auth_code, password, and phone_number are intentionally absent from agent.json and redacted from Agent API responses. The provider credential represented by auth_code is stored encrypted after it is configured. Registration passwords and phone numbers are entered only on the provider page and are not stored by the current QwenPaw workflow. The absence of auth_code from public configuration does not mean that the user did not configure it.

QwenPaw resolves the encrypted provider credential into the managed DriverCard only at runtime. Never read, decrypt, print, copy, or modify credentials.yaml, and never search files or logs for a secret. Use the workflows below.

Workflow: Connect or Create the Mailbox Account

Step 1 — Read public mailbox state

Read mail.is_new_account, mail.credential.name, and mail.credential.domain from agent.json. Treat provider as empty for every currently supported personal domain.

If mail is missing, ask the user to configure Email Management in the QwenPaw Agent Settings UI before continuing.

Step 2a — is_new_account is false: manage an existing mailbox

  1. Call check_auth directly. The managed DriverCard already receives the stored email credential through a runtime credential reference; do not look it up in agent.json and do not call set_credentials merely because secret fields are absent.
  2. If check_auth succeeds, perform the requested mailbox operation.
  3. If credentials are missing or invalid, ask the user to edit this agent in QwenPaw, choose Manage your personal mailbox, enter the mailbox credential again, and save. Retry check_auth after the agent reloads.
  4. If the user explicitly supplies an email and credential in the current conversation, set_credentials may be used as a temporary session-only override, followed by check_auth. It does not update the encrypted QwenPaw configuration and is lost when the MCP process restarts.

For temporary set_credentials, pass the full email and the provider-specific credential in the auth_code parameter. For aliyun.com, that parameter contains the login password; for the other supported domains, it contains the 16-character authorization code or app password.

Step 2b — is_new_account is true: register a dedicated mailbox

Use the public username and domain from agent.json. If the username is blank, let create_mailbox generate one or agree on one with the user.

Registration passwords and phone numbers are entered on the provider page and are deliberately unavailable to the Agent. Do not try to recover them from files. The optional credential field in the QwenPaw dedicated-mailbox form is only for the final provider authorization code, app password, or mailbox login password after registration. Use one of these paths:

Preferred path — visible browser registration
  1. For NetEase or Tencent domains, call create_mailbox(domain, username) first to validate the username and obtain current provider guidance.
  2. Open the provider's official registration page in a visible browser.
  3. When the page requests a password, phone number, CAPTCHA, SMS code, or other identity verification, ask the user to enter it directly in the visible browser. If the user explicitly provides a value for this task, use it only for the current registration and never persist or repeat it.
  4. Keep the browser open while waiting for user action, then continue after confirmation.

Official registration entry points:

DomainRegistration entryNotes
163.com, 126.com, yeah.nethttps://zc.reg.163.com/regInitializedShared NetEase flow; phone verification required
qq.com, foxmail.comhttps://ssl.zc.qq.com/v3/index-chs.htmlQQ registration; phone verification required
sina.comhttps://mail.sina.com.cn/register/weixin.phpWeChat-authorized registration
sina.cnhttps://mail.sina.cn/register/regmail.phpPhone/SMS registration
gmail.comhttps://accounts.google.com/signupEnable 2-Step Verification, then create an app password
aliyun.comUnavailableNew personal registrations are closed; use an existing account

Consider registration successful only after a clear success message or successful inbox access. If the final username differs from the requested one, report the reason and final address.

Fallback path — user-completed registration

For NetEase or Tencent domains, call create_mailbox(domain, username) and relay its alternatives, registration URL, and steps. Ask the user to complete all password, phone, CAPTCHA, and SMS steps in their own browser.

For Sina or Gmail, direct the user to the official entry above. For aliyun.com, explain that a new account cannot be registered and ask the user to choose another supported domain.

Step 3 — After successful registration

  1. Do not write an authorization code, password, or phone number into agent.json.
  2. Ask the user to edit the agent in the QwenPaw Agent Settings UI and keep Provision a dedicated mailbox selected. Enter the final mailbox name and the optional provider credential shown for that domain, then save. The field accepts a 16-character authorization code/app password for NetEase, Tencent, Sina, or Gmail, and a login password for a provider that uses one. QwenPaw will automatically set is_new_account to false, store the secret in encrypted form, synchronize the managed DriverCard, and reload the agent.
  3. After reload, call check_auth. Do not run other mail tools until it succeeds.
  4. Read CONTACTS.md before contact-dependent work.

Available Tools

Read-Only Tools

ToolPurpose
list_foldersList all mailbox folders
list_messagesList message envelopes in a folder with pagination
get_messageFetch one message by folder and UID
get_attachmentGet an attachment by filename or index
search_messagesSearch by keyword, sender, or date range
check_authVerify fresh IMAP and SMTP logins using the current runtime credential
create_mailboxReturn registration guidance for supported NetEase/Tencent domains
list_threadsList conversation threads with incremental sync
search_threadsSearch conversation threads
get_threadGet all messages in one thread
get_mailbox_statsGet recent mailbox statistics

Write Tools

ToolPurpose
send_messageSend a plain-text email with to/cc/bcc
reply_messageReply with proper threading headers
forward_messageForward a message as an RFC 822 attachment
mark_messagesMark messages read/unread/flagged/unflagged
move_messageMove a message to another folder
create_folderCreate a mailbox folder
set_credentialsSet a temporary in-memory credential for this MCP process
clear_credentialsClear the temporary override and fall back to injected startup credentials
update_threadAdd or remove custom thread labels

Destructive Tools

ToolPurpose
delete_messagePermanently delete one message
delete_threadMove a whole thread to trash

Safety and Reliability Notes

  • Never guess or expose an authorization code, password, phone number, CAPTCHA, or SMS code.
  • Never interpret redacted secret fields as empty credentials; verify with check_auth.
  • Never place secrets in agent.json, DriverCard YAML, CONTACTS.md, logs, or chat summaries.
  • Confirm with the user before delete_message or delete_thread.
  • Message UIDs are folder-scoped and can change. Refresh with list_messages or search_messages immediately before acting.
  • After any runtime or UI credential change, call check_auth first.
  • Update CONTACTS.md when the user wants newly discovered contact information retained, but never store credentials there.

Frequently asked questions

What does the Mailbox AI skill do?

Use this skill whenever the user needs ANY mailbox/email operation — checking, reading, searching, sending, replying, forwarding, organizing or deleting email, managing threads, connecting a personal mailbox, or registering a new mailbox. This skill is the single entry point for email tasks and orchestrates qwenpawmail-mcp for nine supported personal-mail domains.

Why use Mailbox on TypingMind?

Because you install it once and use it with any model. Mailbox is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Mailbox in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/agentscope-ai/QwenPaw/tree/main/src/qwenpaw/agents/skills/mailbox-en. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Mailbox?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Mailbox?

As many as you like. As long as a model supports skills, you can use Mailbox with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Mailbox AI skill free?

Yes. It is published on GitHub by agentscope-ai under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇