# Command Mail Learn: full text (English) > All English pages from https://commandmail.app/en/learn as Markdown in one file. Generated 2026-10-17. --- # Best email MCP servers for Gmail, Outlook and IMAP in 2026 > Which email MCP server fits Gmail, Outlook, IMAP or several addresses? Official, self-hosted and hosted servers compared on stated criteria, with prices. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/compare/best-email-mcp-servers · Deutsch: https://commandmail.app/de/vergleich/beste-e-mail-mcp-server.md The right email MCP server depends on where your mail lives and whether your agent should send on its own, so no single server is best for everyone. Google now runs an official Gmail MCP server in preview without a send tool, google_workspace_mcp and ms-365-mcp-server are actively maintained self-hosted options for Google and Microsoft accounts, mcp-email-server covers IMAP mailboxes that allow password sign-in, AgentMail creates inboxes for agents, and Superhuman Mail and Zapier MCP serve their own users[^1][^3][^4][^7][^16][^18][^20]. Command Mail is our product; it fits when several addresses from different providers should share one inbox and you want each sending permission granted separately[^22]. ## How we compared, and our interest Command Mail is our product, so we kept the criteria explicit and applied them to every option: - **Accounts:** Gmail, Google Workspace, Microsoft 365, personal Microsoft accounts, IMAP mailboxes, or new addresses created for agents. - **Sign-in:** OAuth with the provider, or a mailbox password stored by the server. - **Sending and safeguards:** whether the agent can send, and what sits in front of it: a confirmation prompt, a read-only mode, a review window or a daily cap. - **Hosting:** a service run by the vendor, or your own machine or cloud project. - **Maintenance:** for self-hosted projects, GitHub data on 2026-10-17: stars, last push, archive status. - **Price:** taken from each vendor's pricing page on 2026-10-17. We read vendor documentation, READMEs and the GitHub API. We did not run the servers side by side, and where a fact isn't documented we say so. The limits of Command Mail get their own section below. ## Hosted email MCP services compared These four services run on the vendor's infrastructure; you add a URL to Claude, ChatGPT, Cursor or Codex and sign in[^16][^18][^20][^22]. | Criterion | Command Mail | AgentMail | Superhuman Mail | Zapier MCP | |---|---|---|---|---| | Accounts | Any address you own via forwarding, custom domains included; import of old mail[^24] | New inboxes for agents rather than your existing addresses[^16] | Gmail and Outlook accounts set up in Superhuman Mail[^18] | Gmail and other apps connected to your Zapier account[^20] | | Sign-in | OAuth 2.1 with PKCE and dynamic client registration[^22] | OAuth for Claude and ChatGPT, API key for Cursor[^16] | OAuth with your Superhuman account[^18] | Zapier login; Zapier manages app credentials[^20] | | Sending and safeguards | Separate permissions for drafts, sending the exact approved version, and independent sending with a daily limit (50 by default)[^22] | send_message sends directly; allow and block lists; API keys with allow-listed permissions, scoped to an inbox or pod[^16][^25] | send_draft sends instantly with undo window; read-only mode available, per-call confirmation in the client recommended[^18] | Choose tools per server, including a fixed tool set[^20] | | Autonomy for sorting and filing | Per mailbox: read, suggest, rules, auto; does not limit sending[^23] | Not documented | Not documented | Not documented | | Price on 2026-10-17 | Free €0 without sending; Pro €19/month as a founder price for the first 100 paying Pro accounts while the subscription runs, otherwise €29/month[^22] | Free 0 $; Developer 20 $/month[^17] | Business 40 $ monthly or 396 $ a year; no MCP on Starter[^19][^18] | Free 0 $ (100 tasks/month); Professional as the cheapest paid plan, Team from 69 $/month billed annually; two tasks per call[^21][^20] | ## Built-in, official and self-hosted options If you'd rather not add a service, there are three groups: the connectors built into the assistants, servers from Google and Microsoft themselves, and projects you run yourself[^11][^1][^3][^4][^7]. | Option | Accounts | Sign-in | Sending | Hosting | Maintenance on 2026-10-17 | |---|---|---|---|---|---| | Claude Gmail connector | One Google account per connection | Google OAuth | Yes, approval before each send by default[^11] | Anthropic, all Claude plans | Vendor service; attachment metadata only[^11] | | ChatGPT Gmail and Outlook apps | Gmail, Outlook; some apps allow more accounts | Google or Microsoft OAuth | Depends on the app; permission level can be set per connected account[^14] | OpenAI | Vendor service | | Google Gmail MCP (Developer Preview) | Google accounts | Your own OAuth client in a Cloud project | No, drafts only[^1] | Google; preview program required | Reference last updated 2026-07-21[^2] | | google_workspace_mcp | Google Workspace and Gmail | Your own OAuth client; OAuth 2.1 multi-user | Yes; read-only mode and tool tiers[^3] | Self-hosted, local or central | 3,353 stars, pushed 2026-10-09, MIT[^3] | | ms-365-mcp-server | Personal Microsoft accounts; work accounts with --org-mode | Softeria's default app or your own Azure app | Yes; read-only mode and --allowed-scopes[^4] | Self-hosted, local or HTTP | 1,052 stars, pushed 2026-10-17, MIT[^4] | | outlook-mcp (ryaker) | Outlook via Microsoft Graph | Your own Azure app with client secret | Yes, send-email tool[^6] | Self-hosted, local | 445 stars, pushed 2026-03-30, no license stated[^6] | | mcp-email-server | IMAP/SMTP accounts whose provider allows password or app-password sign-in; several at once | Email address and password | Yes, via SMTP[^7] | Self-hosted on Windows, macOS, Linux | 361 stars, pushed 2026-10-13, BSD-3-Clause[^7] | | thunderbird-mcp and apple-mail-mcp | Whatever your desktop mail app has | Through the mail app | Thunderbird opens a review window by default; Apple Mail can run --read-only[^8][^9] | Your computer | 384 and 230 stars, pushed 2026-10-05 and 2026-10-16, both MIT[^8][^9] | | Gmail-MCP-Server (GongRzhe) | Gmail | Not checked | Yes | Self-hosted | Archived, last push 2025-08-06[^10] | ## Each option in more detail ### Claude and ChatGPT connectors Claude's Gmail connector searches, reads, drafts, sends and forwards for one Google account and asks before each send by default[^11]. Its separate Microsoft 365 connector accepts work accounts in an Entra tenant only and sends only after an admin enables write tools[^12]. ChatGPT's Gmail and Outlook apps follow a permission setting that ranges from Always ask to Allow all actions[^14]. Good for: one account you use inside a conversation, with nothing to install. ### Google's official Gmail MCP server Google offers a remote MCP server at gmailmcp.googleapis.com as part of its Workspace Developer Preview Program[^1]. Setup means enabling the Gmail API and the Gmail MCP API in a Cloud project, configuring an OAuth consent screen with the gmail.readonly and gmail.compose scopes, and creating a client flagged for AI agents. The tools search threads, read messages, manage labels and create drafts, and you send from Gmail yourself[^2]. Google also tells developers to screen prompts and responses for injection, for example with Model Armor. Good for: teams comfortable with Google Cloud who want reading and drafting and prefer that the agent can't send. ### google_workspace_mcp taylorwilsdon's project has 3,353 GitHub stars, the most of the self-hosted servers we looked at[^3]. Its README lists more than 120 tools across twelve Google services, three tool tiers, a read-only mode, per-service permissions and OAuth 2.1 for central multi-user deployment. You bring your own OAuth client from the Google Cloud Console, and the README says data goes nowhere except Google's APIs. Good for: developers and IT teams who want Gmail alongside Calendar, Drive and Docs under their own control. ### ms-365-mcp-server, outlook-mcp and Microsoft's own servers Softeria's ms-365-mcp-server uses the Graph API, works with personal Microsoft accounts by default and with work accounts and shared mailboxes in --org-mode[^4]. The --read-only flag and --allowed-scopes keep both the tool list and the requested permissions narrow. For organizations, Microsoft provides Work IQ MCP through Agent 365 and steers new projects away from its older workload-specific Mail server[^5]. The smaller outlook-mcp by ryaker needs your own Azure app and secret, and its last push was on 2026-03-30[^6]. Good for: ms-365-mcp-server for individuals and admins on Microsoft accounts; Work IQ for companies that manage Microsoft 365 centrally. ### mcp-email-server for IMAP mcp-email-server reads, searches, organizes and sends over IMAP and SMTP, handles several accounts and runs on Windows, macOS and Linux[^7]. Each account is set up with its address and password, so the provider has to allow IMAP and SMTP with that kind of sign-in; with two-factor sign-in it usually means an app password. It provides a direct route for mailboxes that have no API of their own. Good for: developers who want a local connection to an IMAP provider that allows password sign-in and who accept a stored password. ### Desktop mail apps: Thunderbird and Apple Mail thunderbird-mcp exposes 40 tools through a Thunderbird extension that listens on localhost only[^8]. Send, reply and forward open a review window by default, and direct sending only works after you switch that safety setting off yourself. apple-mail-mcp works with Apple Mail on macOS, and its --read-only flag blocks every send[^9]. Good for: people whose accounts already live in a desktop client and who want the agent to stay on their own machine. ### AgentMail AgentMail gives agents their own inboxes through an API and runs a hosted MCP server with 43 tools, including sending, drafts, attachments and allow and block lists[^16]. Claude and ChatGPT connect with OAuth, while Cursor and Windsurf use an API key. An API key can be scoped to one inbox or a group of inboxes (a pod) and limited to specific permissions such as reading, drafting or sending[^25]. Agent Armor scans inbound mail for attempts to manipulate agents and can hold messages; AgentMail lists it as a beta in its Labs, available in US production but not yet in the EU[^26]. The Free plan costs 0 $ with three inboxes, and the Developer plan is 20 $ a month[^17]. Good for: developers whose agent needs an address of its own, for sign-ups, support bots or outbound workflows. ### Superhuman Mail MCP Superhuman's MCP server connects the Gmail and Outlook accounts in your Superhuman Mail seat, several at once, to Claude, ChatGPT, Codex, Cursor, Gemini CLI, Perplexity and other clients[^18]. It searches, drafts in your voice, manages events and sends with send_draft; because that send is immediate, Superhuman advises turning on per-call confirmation in your AI client, and you can run the server in read-only mode. You need the Business plan at 40 $ monthly or 396 $ a year, since the Starter plan doesn't include the MCP server[^19]. Good for: Superhuman customers who want to run their existing inbox from an AI client. ### Zapier MCP Zapier MCP links an AI client to more than 9,000 apps, Gmail among them, and handles app connections, credentials and rate limits[^20]. Zapier MCP is available on every plan. Each successful tool call uses two tasks from your plan, so the 100 tasks in the Free plan cover about 50 calls. The cheapest paid plan is Professional, and Team starts at 69 $ a month billed annually[^21]. You can let the agent enable tools as it goes or pin a fixed tool set to keep access tight. Good for: workflows that tie email to a CRM, spreadsheets or chat. ### Archived: GongRzhe's Gmail MCP server GongRzhe's server still shows up in search suggestions, but the GitHub repository is archived and the last push was on 2025-08-06[^10]. Bugs and security issues there won't be fixed. Good for: existing installs that you should review for support and security risks; for new production setups we recommend an actively maintained project. ## When another server is the better choice Several options beat Command Mail in specific, nameable situations: - **You need changes in the original mailbox.** google_workspace_mcp, ms-365-mcp-server and the Claude and ChatGPT connectors work on the real Gmail or Outlook account, so labels, folders and drafts stay in sync[^3][^4][^11]. Command Mail works on forwarded copies. - **You want nothing in between.** Self-hosted servers run on your machine or in your own cloud project; google_workspace_mcp's README says it sends data only to Google's APIs[^3][^7]. Command Mail is a hosted service and its source code is not public. - **You want no license fee.** The MIT- and BSD-licensed self-hosted projects charge nothing for the software; you still pay for hosting and possibly a provider account or AI client. mcp-email-server reaches IMAP mailboxes whose provider allows password or app-password sign-in[^7]. outlook-mcp states no license, so check the terms of use before relying on it[^6]. - **You want narrowly scoped keys or read-only access.** AgentMail limits API keys to specific inboxes and permissions, and Superhuman Mail and several self-hosted projects offer a read-only mode[^25][^18][^3][^4]. - **You need many agent addresses.** AgentMail is built for agents that each need their own inbox, at volume[^16]. - **You want a polished mail client too.** Superhuman Mail comes with its own client, calendar, reminders and read status for sent mail[^18]. - **Email is one step among many apps.** Zapier connects Gmail with thousands of other apps[^20]. ## When Command Mail fits Command Mail fits when Gmail, Outlook and custom-domain addresses should land in one inbox and an agent should work there under fixed rules[^24]. Each address gets its own private forwarding destination; with the mailboxes permission an agent can set that up itself, verify a domain through a TXT record and import old mail from MBOX or EML files. Sending comes in three permissions that you grant separately[^22]. mail:draft only saves drafts. mail:send-approved sends exactly the draft version a person approved in the browser, once. mail:send allows independent sending with a unique send key per email, a daily limit of 50 by default and at most 20 recipients per message. For sorting and filing, each mailbox has an autonomy level of read, suggest, rules or auto[^23]. These levels do not limit sending: sending permissions apply to the whole account, so an agent with mail:send can send from a mailbox set to read. Don't give mail:send to an agent that should only sort. Command Mail treats message content as untrusted data and flags suspected prompt injection. The reply tool refuses to answer messages whose headers mark them as automated, such as Auto-Submitted, Precedence: bulk or List-Id[^24]. Claude, ChatGPT, Codex, Cursor and your own scripts over the CLI or REST all get the same permissions and limits. ## Who Command Mail is not for - People who need edits to sync back to Gmail or Outlook: forwarding and import are not synchronization, and nothing flows back to the original mailbox[^24]. - People who want to send for free: the Free plan reads, sorts and drafts but doesn't send; sending starts with Pro[^22]. - Teams that must self-host or audit the code: Command Mail is a hosted service run by FYN Labs LLC in the United States. - ChatGPT users who want a one-click directory install today: our listing is still under review. Until then the route is a custom MCP server, which OpenAI allows on the web: read and fetch actions only on Pro, and full support including sending on Business, Enterprise and Edu, subject to roles[^15]. ## Setup: connecting an email MCP server Hosted servers follow the same pattern almost everywhere. In Claude, open Customize → Connectors and add a custom connector with the server URL; the Free plan allows one custom connector[^13]. In ChatGPT, go to Plugins, select the plus button and choose Add custom MCP server; you no longer need to turn on developer mode first. This works on the web only, with read and fetch actions on Pro and full support on Business, Enterprise and Edu if your role allows creating connections[^15]. Then sign in with the vendor and approve the requested permissions. For Command Mail, these are the commands for Claude Code and the Codex CLI[^22]: ```bash claude mcp add --transport http commandmail https://commandmail.app/mcp codex mcp add commandmail --url https://commandmail.app/mcp codex mcp login commandmail ``` Command Mail is also listed in Claude's connectors directory. On first sign-in, approve reading and drafting only and add sending later as its own permission[^23]. Every session starts with the capabilities call, which returns the owner, granted permissions, mailboxes and limits. Self-hosted servers add their own steps: an OAuth client in a Google Cloud project, an Azure app registration or an app password. Each project's README walks through them[^3][^4][^7]. ## Questions to ask any email MCP server Email is a risky channel for agents because every incoming message is text written by strangers. Google's own Gmail server guide warns that hidden instructions in mail can hijack a session and lead to data being changed, stolen or deleted[^1]. Three questions help with any choice: can sending be switched off separately from reading, does the server see your password or only a token, and who runs it and how recently was it updated? For how ChatGPT and Claude confirm sends, read [Can ChatGPT send emails for you?](https://commandmail.app/en/learn/can-chatgpt-send-emails). To bring several accounts to one agent, follow [Use multiple email accounts with an AI agent](https://commandmail.app/en/guides/multiple-email-accounts-ai-agent); the built-in Gmail route is in [Connect Gmail to Claude](https://commandmail.app/en/guides/connect-gmail-to-claude). Command Mail's permissions and connection options are on [For agents](https://commandmail.app/en/agents) and in the [documentation](https://commandmail.app/en/docs). You can [start free](https://commandmail.app/login?signup=1) and connect your agent afterwards. ## Frequently asked questions ### Is there an official Gmail MCP server from Google? Yes, in Developer Preview at gmailmcp.googleapis.com. It requires membership in the Google Workspace Developer Preview Program, a Google Cloud project and your own OAuth client. It searches, reads, labels and creates drafts, and it has no send tool. ### Which email MCP server works with IMAP accounts? mcp-email-server talks IMAP and SMTP, handles several accounts and runs locally on Windows, macOS and Linux. Its README sets accounts up with an email address and password, which works with providers that allow IMAP and SMTP and sign-in with a password or app password. Command Mail accepts any address by forwarding instead, so the agent never holds a mailbox password. ### Can I still use GongRzhe's Gmail MCP server? The repository is archived on GitHub and its last push was on 2025-08-06. For a new setup, a maintained project such as google_workspace_mcp or Google's own server is the safer choice. ### Which email MCP servers can send without asking? Most can, if the client allows it. Superhuman's send_draft sends instantly; Superhuman recommends per-call confirmation in your AI client and also offers a read-only mode. AgentMail lets you issue API keys without send permission. Thunderbird's MCP opens a review window by default. Command Mail splits drafting, approved sending and independent sending into separate permissions with a daily limit. ### Do any hosted email MCP services have a free plan? AgentMail, Zapier and Command Mail start free. Command Mail Free does not send, AgentMail Free includes three agent inboxes, and every successful Zapier MCP tool call uses two tasks from your plan. ## Sources 1. [Google for Developers: Configure the Gmail MCP server](https://developers.google.com/workspace/gmail/api/guides/configure-mcp-server), accessed 2026-10-17 2. [Google for Developers: MCP Reference gmailmcp.googleapis.com](https://developers.google.com/workspace/gmail/api/reference/mcp), accessed 2026-10-17 3. [GitHub taylorwilsdon/google_workspace_mcp (README, stars, last push)](https://github.com/taylorwilsdon/google_workspace_mcp), accessed 2026-10-17 4. [GitHub Softeria/ms-365-mcp-server (README, stars, last push)](https://github.com/Softeria/ms-365-mcp-server), accessed 2026-10-17 5. [Microsoft Learn: Understand MCP server options in Agent 365](https://learn.microsoft.com/en-us/microsoft-agent-365/tooling-servers-overview), accessed 2026-10-17 6. [GitHub ryaker/outlook-mcp (README, stars, last push)](https://github.com/ryaker/outlook-mcp), accessed 2026-10-17 7. [GitHub Wh1isper/mcp-email-server (README, stars, last push)](https://github.com/Wh1isper/mcp-email-server), accessed 2026-10-17 8. [GitHub TKasperczyk/thunderbird-mcp (README, stars, last push)](https://github.com/TKasperczyk/thunderbird-mcp), accessed 2026-10-17 9. [GitHub patrickfreyer/apple-mail-mcp (README, stars, last push)](https://github.com/patrickfreyer/apple-mail-mcp), accessed 2026-10-17 10. [GitHub GongRzhe/Gmail-MCP-Server (archived)](https://github.com/GongRzhe/Gmail-MCP-Server), accessed 2026-10-17 11. [Claude Help Center: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 12. [Claude Help Center: Connect to Microsoft 365](https://support.claude.com/en/articles/15183774-connect-to-microsoft-365), accessed 2026-10-17 13. [Claude Help Center: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 14. [OpenAI Help Center: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 15. [OpenAI Help Center: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 16. [AgentMail Docs: MCP](https://docs.agentmail.to/integrations/mcp), accessed 2026-10-17 17. [AgentMail: Pricing](https://www.agentmail.to/pricing), accessed 2026-10-17 18. [Superhuman Help Center: Superhuman Mail MCP Server](https://help.superhuman.com/hc/en-us/articles/46005696690317-Superhuman-Mail-MCP-Server), accessed 2026-10-17 19. [Superhuman Help Center: Pricing Plans](https://help.superhuman.com/hc/en-us/articles/46178817646861-Pricing-Plans), accessed 2026-10-17 20. [Zapier Help Center: What is Zapier MCP?](https://help.zapier.com/hc/en-us/articles/48308034391821-What-is-Zapier-MCP), accessed 2026-10-17 21. [Zapier: Pricing](https://zapier.com/pricing), accessed 2026-10-17 22. [Command Mail: llms.txt with plans, MCP endpoint and permissions](https://commandmail.app/llms.txt), accessed 2026-10-17 23. [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 24. [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 25. [AgentMail Docs: Permissions](https://docs.agentmail.to/permissions), accessed 2026-10-17 26. [AgentMail Docs: Agent Armor (Labs, Beta)](https://docs.agentmail.to/agent-armor), accessed 2026-10-17 [^1]: [Google for Developers: Configure the Gmail MCP server](https://developers.google.com/workspace/gmail/api/guides/configure-mcp-server), accessed 2026-10-17 [^2]: [Google for Developers: MCP Reference gmailmcp.googleapis.com](https://developers.google.com/workspace/gmail/api/reference/mcp), accessed 2026-10-17 [^3]: [GitHub taylorwilsdon/google_workspace_mcp (README, stars, last push)](https://github.com/taylorwilsdon/google_workspace_mcp), accessed 2026-10-17 [^4]: [GitHub Softeria/ms-365-mcp-server (README, stars, last push)](https://github.com/Softeria/ms-365-mcp-server), accessed 2026-10-17 [^5]: [Microsoft Learn: Understand MCP server options in Agent 365](https://learn.microsoft.com/en-us/microsoft-agent-365/tooling-servers-overview), accessed 2026-10-17 [^6]: [GitHub ryaker/outlook-mcp (README, stars, last push)](https://github.com/ryaker/outlook-mcp), accessed 2026-10-17 [^7]: [GitHub Wh1isper/mcp-email-server (README, stars, last push)](https://github.com/Wh1isper/mcp-email-server), accessed 2026-10-17 [^8]: [GitHub TKasperczyk/thunderbird-mcp (README, stars, last push)](https://github.com/TKasperczyk/thunderbird-mcp), accessed 2026-10-17 [^9]: [GitHub patrickfreyer/apple-mail-mcp (README, stars, last push)](https://github.com/patrickfreyer/apple-mail-mcp), accessed 2026-10-17 [^10]: [GitHub GongRzhe/Gmail-MCP-Server (archived)](https://github.com/GongRzhe/Gmail-MCP-Server), accessed 2026-10-17 [^11]: [Claude Help Center: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 [^12]: [Claude Help Center: Connect to Microsoft 365](https://support.claude.com/en/articles/15183774-connect-to-microsoft-365), accessed 2026-10-17 [^13]: [Claude Help Center: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 [^14]: [OpenAI Help Center: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 [^15]: [OpenAI Help Center: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 [^16]: [AgentMail Docs: MCP](https://docs.agentmail.to/integrations/mcp), accessed 2026-10-17 [^17]: [AgentMail: Pricing](https://www.agentmail.to/pricing), accessed 2026-10-17 [^18]: [Superhuman Help Center: Superhuman Mail MCP Server](https://help.superhuman.com/hc/en-us/articles/46005696690317-Superhuman-Mail-MCP-Server), accessed 2026-10-17 [^19]: [Superhuman Help Center: Pricing Plans](https://help.superhuman.com/hc/en-us/articles/46178817646861-Pricing-Plans), accessed 2026-10-17 [^20]: [Zapier Help Center: What is Zapier MCP?](https://help.zapier.com/hc/en-us/articles/48308034391821-What-is-Zapier-MCP), accessed 2026-10-17 [^21]: [Zapier: Pricing](https://zapier.com/pricing), accessed 2026-10-17 [^22]: [Command Mail: llms.txt with plans, MCP endpoint and permissions](https://commandmail.app/llms.txt), accessed 2026-10-17 [^23]: [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 [^24]: [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 [^25]: [AgentMail Docs: Permissions](https://docs.agentmail.to/permissions), accessed 2026-10-17 [^26]: [AgentMail Docs: Agent Armor (Labs, Beta)](https://docs.agentmail.to/agent-armor), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Can ChatGPT send emails for you? And can Claude? > Yes, if the connected mail app offers a send action, usually after you approve. How ChatGPT and Claude confirm sends, where limits are and how to keep control. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/learn/can-chatgpt-send-emails · Deutsch: https://commandmail.app/de/wissen/kann-chatgpt-e-mails-senden.md Yes. ChatGPT can send email once you connect a mail account such as Gmail or Outlook and that app offers a send action. OpenAI treats sending an email as an action that may need extra review, so depending on your permission setting ChatGPT asks first or refuses[^1]. Claude can send, reply and forward through its Gmail connector, and it asks for your approval before each of those actions by default[^7]. In ChatGPT, whether a send needs your confirmation depends on the app, the connected account and your permission settings. How much control you keep beyond that prompt depends on the setting you pick, which accounts you connect and whether an agent is supposed to work while you are away. ## How sending works in ChatGPT ChatGPT sends mail through connected apps, and a permission setting decides when it stops to ask. On an eligible personal account you set the default under Settings → Plugins → Permissions; if your Settings show Apps instead of Plugins, check the controls offered there. You can also give a single connected account its own setting, different from other accounts on the same app. Depending on your account and app, there are up to four options[^1]: - **Always ask:** ChatGPT asks before reading app data and before making changes. - **Allow read actions:** it reads freely but asks before changes such as a send. - **Allow low-risk actions:** low-risk actions run on their own; higher-risk ones need confirmation or are denied. - **Allow all actions:** supported actions run without further prompts. OpenAI calls this elevated risk and does not offer it in the account-wide or workspace-wide default. When an action needs approval, ChatGPT shows a card naming the app, the account and the proposed step. You can deny it, allow it once, allow low-risk actions for that account, or, for an eligible personally connected account, always allow. Members of managed workspaces don't get the persistent option[^1]. What the email apps can do differs by provider. For Outlook, OpenAI lists the Microsoft permission Mail.Send and describes sending plain-text email from or on behalf of shared and delegated mailboxes[^2]. For Gmail, ChatGPT requests the Google scope gmail.modify[^3], which Google describes as permission to read, compose and send email, short of permanent deletion[^4]. That scope alone doesn't prove a send tool is available on your account: the OpenAI articles we checked describe a concrete send action for Outlook, but not for Gmail. Which actions you actually see still depends on the app, your plan, your region and your workspace settings. ## Can Claude send emails? Yes. Claude's Gmail connector searches and reads mail, writes drafts, sends, replies, forwards, and manages labels and threads[^7]. Claude asks before sending, replying or forwarding unless an owner on a Team or Enterprise plan has allowed members to skip the prompt. The Microsoft 365 connector is available on every Claude plan but only for work accounts in a Microsoft Entra tenant. Personal addresses at outlook.com, hotmail.com or live.com are not supported[^8]. Claude can send from Outlook there only if an admin has turned on write tools. Messages Claude sends carry a header that marks them as agent-initiated, and the write tools can't add attachments yet. ## Where the built-in options run out For a single Gmail or work Outlook account, the built-in connectors cover a lot. The friction shows up in a few predictable places: - **One account per connection.** Claude works with the one Google account you connected[^7], and a request for multiple Gmail accounts has been open since March 2026[^9]. ChatGPT supports extra accounts for some apps, though not all[^5]; for Codex, users have asked for multiple named Gmail accounts in an open issue[^12]. - **Attachments.** Claude sees Gmail attachment metadata, not the file contents[^7]. - **Success messages you can't trust blindly.** One open report describes a scheduled Claude routine that reported success although its Gmail send_message call did nothing[^10]. For Outlook, OpenAI tells users to check the mailbox and sent items before repeating a send whose outcome is unclear[^2]. - **Forward means send.** Another open request asks for Claude's forward action to create a draft instead of sending right away[^11]. - **Approval is per action.** ChatGPT lets each connected account have its own permission level[^1], but neither provider's help pages document a daily sending cap you set yourself. ## What happens to the content of your mail The Gmail and Outlook connections described here get read access to your mailbox along with the ability to send. The provider's sign-in screen shows which permissions a connection actually requests, so read it before you agree. Anthropic says it does not train its models on Gmail, Drive or Calendar connector data, and retrieved data stays with the chat it was used in until you delete that chat[^7]. It names one exception for Free, Pro and Max: if you allow your chats to be used for training, content you copy from these services and Claude responses that include specific information from them may be used to improve its models. OpenAI says it does not train its generalized models on data accessed directly from connected Google apps. The exceptions are conversations you submit as feedback, content you copy, paste or upload yourself, and content included in ChatGPT's response. With "Improve the model for everyone" switched off, OpenAI doesn't use that app data either[^3]. For other connected apps on Free, Plus, Go and Pro, the same setting decides[^6]. ## Practical ways to stay in control These habits help with any assistant: 1. **Start with confirmation on.** Keep Claude's default prompt and set ChatGPT to Always ask or Allow read actions. 2. **Ask for drafts.** Phrase requests as "draft a reply, but don't send it". Anthropic uses exactly that wording as an example for its Microsoft 365 connector[^8]. 3. **Read the approval card.** Check recipient, account and text before you approve; an approval covers the step shown on the card. 4. **Check Sent before retrying.** After an error or a timeout, look in the sent folder first. 5. **Treat incoming mail as untrusted text.** A message can carry hidden instructions. Connect only the accounts an agent needs and add send rights after reading and drafting work well. ## When a dedicated agent inbox makes sense A separate tool earns its place when you juggle several addresses, when an agent should run on a schedule without you, or when several agents should follow the same sending rules. Command Mail is our product, and it is built for those cases[^14]. Control in Command Mail comes from individual permissions you approve when you connect an agent[^15]. With mail:draft the agent saves drafts and never sends. With mail:send-approved it sends exactly the draft version a person approved, once; if the text changes afterwards, it needs a new approval. Only mail:send lets an agent send on its own, with a unique send key per email, a daily limit of 50 by default and at most 20 recipients per message. Sending without approval starts switched off. For sorting, filing and rules, each mailbox also has an autonomy level: read, suggest, rules or auto, and new mailboxes start on auto. That level does not limit sending. Sending permissions apply to the whole account, so an agent with mail:send can send from a mailbox set to read. Don't give mail:send to an agent that should only sort. Command Mail treats message content as untrusted data and flags suspected prompt injection. If an agent uses the reply tool on a message whose headers mark it as automated, such as Auto-Submitted, Precedence: bulk or List-Id, Command Mail refuses the reply. The same rules apply whether the agent is Claude, ChatGPT, Codex, Cursor or another MCP client. Mail reaches Command Mail as forwarded copies, and changes don't sync back to the original mailbox. The Free plan doesn't send; sending starts with Pro at €19 a month as a founder price for the first 100 paying Pro accounts and €29 after that[^15]. In Claude you connect Command Mail from the connectors directory. In ChatGPT our directory listing is still under review, so for now you add it under Plugins as a custom MCP server, without turning on developer mode first. OpenAI allows that on the web only: Pro with read and fetch actions, and Business, Enterprise and Edu with full support including write actions, subject to roles and workspace settings[^13]. Sending from ChatGPT therefore needs one of those three plans. The [For agents](https://commandmail.app/en/agents) page and the [documentation](https://commandmail.app/en/docs) cover permissions and connection options in detail. To give one agent all your addresses, follow [Use multiple email accounts with an AI agent](https://commandmail.app/en/guides/multiple-email-accounts-ai-agent). For the built-in route, see [Connect Gmail to Claude](https://commandmail.app/en/guides/connect-gmail-to-claude). Other ways to give an agent email access are compared in [Best email MCP servers for Gmail, Outlook and IMAP](https://commandmail.app/en/compare/best-email-mcp-servers). You can [start free](https://commandmail.app/login?signup=1) and connect your agent with exactly the permissions you want it to have. ## Frequently asked questions ### Will ChatGPT send an email without asking me? Only if you let it. On Always ask or Allow read actions, ChatGPT asks before any change. Allow all actions lets supported actions run without another prompt; OpenAI labels it elevated risk and keeps it out of the account-wide default selector. Each connected account can have its own setting. ### Does Claude ask before it sends an email? Yes. With the Gmail connector, Claude asks for approval before each send, reply and forward by default. On Team and Enterprise plans, owners decide whether members may switch that prompt off. ### Can Claude use my personal Outlook.com address? Not through the Microsoft 365 connector. Anthropic requires a work account in a Microsoft Entra tenant and does not accept @outlook.com, @hotmail.com or @live.com addresses. ### Can ChatGPT or Claude attach files when they send? Claude's Microsoft 365 write tools can't send, forward or draft emails with attachments yet, and its Gmail connector reads attachment metadata only. OpenAI's Outlook article covers reading attachments in your own mailbox but says nothing about attaching files to outgoing mail. ### How do I connect Command Mail to ChatGPT today? Our ChatGPT directory listing is still under review. Until it is live, you add Command Mail as a custom MCP server under Plugins, with no need to turn on developer mode first. OpenAI allows this on the web only: Pro gets read and fetch actions, while Business, Enterprise and Edu get full support including write actions such as sending, subject to roles and workspace settings. In Claude, Command Mail is in the connectors directory. ## Sources 1. [OpenAI Help Center: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 2. [OpenAI Help Center: Outlook Email and Calendar apps in ChatGPT](https://help.openai.com/en/articles/12512241-outlook-email-and-calendar-apps-in-chatgpt), accessed 2026-10-17 3. [OpenAI Help Center: Google app data controls FAQ](https://help.openai.com/en/articles/10408842-google-app-data-controls-faq), accessed 2026-10-17 4. [Google for Developers: Choose Gmail API scopes](https://developers.google.com/workspace/gmail/api/auth/scopes), accessed 2026-10-17 5. [OpenAI Help Center: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 6. [OpenAI Help Center: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 7. [Claude Help Center: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 8. [Claude Help Center: Connect to Microsoft 365](https://support.claude.com/en/articles/15183774-connect-to-microsoft-365), accessed 2026-10-17 9. [GitHub anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 10. [GitHub anthropics/claude-code #97895: Gmail send_message silently no-ops but run reports SUCCEEDED](https://github.com/anthropics/claude-code/issues/97895), accessed 2026-10-17 11. [GitHub anthropics/claude-code #100177: Gmail connector, let forward create a draft instead of sending](https://github.com/anthropics/claude-code/issues/100177), accessed 2026-10-17 12. [GitHub openai/codex #30418: Gmail connector should support multiple named accounts](https://github.com/openai/codex/issues/30418), accessed 2026-10-17 13. [OpenAI Help Center: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 14. [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 15. [Command Mail: llms.txt with plans and permissions](https://commandmail.app/llms.txt), accessed 2026-10-17 [^1]: [OpenAI Help Center: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 [^2]: [OpenAI Help Center: Outlook Email and Calendar apps in ChatGPT](https://help.openai.com/en/articles/12512241-outlook-email-and-calendar-apps-in-chatgpt), accessed 2026-10-17 [^3]: [OpenAI Help Center: Google app data controls FAQ](https://help.openai.com/en/articles/10408842-google-app-data-controls-faq), accessed 2026-10-17 [^4]: [Google for Developers: Choose Gmail API scopes](https://developers.google.com/workspace/gmail/api/auth/scopes), accessed 2026-10-17 [^5]: [OpenAI Help Center: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 [^6]: [OpenAI Help Center: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 [^7]: [Claude Help Center: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 [^8]: [Claude Help Center: Connect to Microsoft 365](https://support.claude.com/en/articles/15183774-connect-to-microsoft-365), accessed 2026-10-17 [^9]: [GitHub anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 [^10]: [GitHub anthropics/claude-code #97895: Gmail send_message silently no-ops but run reports SUCCEEDED](https://github.com/anthropics/claude-code/issues/97895), accessed 2026-10-17 [^11]: [GitHub anthropics/claude-code #100177: Gmail connector, let forward create a draft instead of sending](https://github.com/anthropics/claude-code/issues/100177), accessed 2026-10-17 [^12]: [GitHub openai/codex #30418: Gmail connector should support multiple named accounts](https://github.com/openai/codex/issues/30418), accessed 2026-10-17 [^13]: [OpenAI Help Center: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 [^14]: [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 [^15]: [Command Mail: llms.txt with plans and permissions](https://commandmail.app/llms.txt), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Connect Gmail to ChatGPT: setup, accounts and permissions > Connect Gmail to ChatGPT directly or through Command Mail. Follow the setup steps, check plan and permission limits, and verify access without sending mail. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/guides/connect-gmail-to-chatgpt · Deutsch: https://commandmail.app/de/anleitungen/gmail-mit-chatgpt-verbinden.md To connect Gmail to ChatGPT, open Settings → Plugins, choose the Gmail app and authorize the Google account you want to use. You can then ask questions using the Gmail capabilities available to your account; availability depends on the app, plan and workspace, among other factors.[^1] Use Command Mail when you want a shared workspace for forwarded copies from different providers, after checking ChatGPT's separate custom MCP requirements.[^4][^7] Command Mail is our product. For someone who only needs to find and summarize existing Gmail messages, the built-in connection is the first route to try. This guide follows current vendor instructions and checked Command Mail source code; it is not a claim that we completed a new-account test on every ChatGPT plan. ## Decide where the agent should work The important choice is whether ChatGPT should use your Google account directly or a separate collection of copied messages. A direct connection avoids setting up forwarding and keeps the original mailbox as the source. An external inbox adds another place to manage, which should solve a real requirement rather than become an extra setup step.[^1][^7] For example, a consultant might want one agent to review messages sent to a personal Gmail address and a separate business-domain address. Command Mail can receive both through forwarding. Someone who needs Gmail labels and archive changes reflected in the original account should choose a direct integration with those supported capabilities instead: Command Mail does not sync its changes back.[^7] | Setup question | Built-in Gmail connection | Gmail through Command Mail | |---|---|---| | Source of messages | Your authorized Google account | Forwarded copies; optional historical import | | Extra inbox required | No | Yes | | Multiple accounts | Depends on app and account; check the account controls | Addresses from different providers, subject to plan limits | | Custom MCP setup | Not required for the built-in app | Pro for read/fetch; Business, Enterprise or Edu for full MCP support | | Sending | Depends on the actions the actual app exposes | Pro or above, a separately send-enabled sender address and ChatGPT write access; Gmail forwarding alone does not authorize sending from Gmail | | Changes back in Gmail | Uses the source account | No reverse synchronization | Sources: OpenAI's app, account and MCP guides[^1][^2][^4] and Command Mail's documentation.[^7][^8] A Google OAuth permission is not proof that a particular send tool is available in ChatGPT. ## Prerequisites For a direct connection, you need a Google account with the messages you want and access to the Gmail app in ChatGPT. Availability can vary by plan, region, workspace, role, model and interface. Check the active ChatGPT workspace before starting, especially if you switch between personal and company work.[^1] Your Google account and your ChatGPT account are separate. Connecting a personal Google account does not make work-only information accessible. Your Google Workspace administrator may also need to approve access, even when a ChatGPT administrator has already enabled the app.[^2] For the Command Mail route, you need a Command Mail account, permission to configure Gmail forwarding and an eligible ChatGPT custom MCP connection. Command Mail Free has one mailbox, the last 90 days and no sending. Command Mail Pro includes up to ten mailboxes, full history and sending. ChatGPT Pro is a different subscription: OpenAI limits its custom MCP connections to read/fetch, so paying for Command Mail Pro does not by itself enable sending from ChatGPT.[^4][^8] Custom MCP setup currently takes place on ChatGPT web. On Business, only administrators and owners can create and publish workspace connections. Enterprise and Edu administrators can grant selected members creation rights using role-based access control. Confirm that you have the appropriate role before spending time on forwarding or importing mail.[^4] ## Steps: connect the built-in Gmail app 1. Open ChatGPT and confirm the account and workspace where you want Gmail available. 2. Go to **Settings → Plugins** and select Gmail or the plugin containing the Gmail app. Very old mobile app versions may still show **Apps**.[^1] 3. Choose **Install plugin** if shown, then **Connect** or the setup option offered for that app.[^2] 4. Sign in to the intended Google account. Read the address in the account picker rather than accepting whichever account is already selected. 5. Review the requested services and permissions. If the flow includes more Google apps than you need, use **Select fewer apps** to reduce the selection. All requested permissions are still required for the apps you keep.[^2] 6. Finish the provider authorization, return to ChatGPT and verify the connected account. 7. Start a supported conversation and select Gmail with **@** or the **+** menu. Some interfaces put further choices under **More**.[^1] For a first request, use a message you already know: “Find the latest Gmail message with the subject Workshop schedule. Report the sender, date and outstanding questions. Do not change or send anything.” This gives you an answer you can check without asking the agent to reorganize an inbox. App selection applies when you request new information. If a later message needs another lookup, select or mention the app again as needed. Discussing information already returned is different from asking for a fresh mailbox search; make that distinction clear in your request.[^4] ## Set approval behavior before using broader tasks Choose **Allow read actions** if you want reads to proceed while changes require a prompt. **Always ask** also prompts before reading. **Allow low-risk actions** can approve low-risk operations automatically, while higher-risk operations may need confirmation or be denied. Allow low-risk actions applies unless an account, workspace, app or connection policy overrides it. **Allow all actions** permits supported actions without another approval prompt and carries elevated risk; it is not offered in the standard account-wide or workspace-wide selector.[^3] For an eligible personal connection, open **Settings → Plugins → Gmail**. If **Connected accounts** appears, open the intended account's **••• → Settings → Permissions**, choose an available option and select **Save** if shown. With a single account, use its available Permissions control. Managed-workspace members may instead see controls governed by their organization.[^3] Approval settings do not grant a missing capability or revoke Google permissions. They control when ChatGPT asks before using access already available. If you need to stop future access, disconnect the account. Different connections to the same app can have different approval settings, so review the particular account rather than relying only on a global default.[^3] ## Steps: connect forwarded Gmail through Command Mail Command Mail uses forwarding, not Gmail account access. Signing in with Google identifies you to Command Mail; it does not give the service access to Gmail, Drive or Calendar. You configure the flow of messages separately.[^7] 1. Sign in to Command Mail, connect the Gmail address under the mailbox settings and copy its private inbound address. 2. In Gmail on a computer, open **Settings → See all settings → Forwarding and POP/IMAP**, or the **Forwarding** tab if that is what your interface shows. Select **Add a forwarding address** and enter the private inbound address. Confirm with **Next → Proceed → OK**.[^6] 3. Open the verification email in Command Mail and follow its confirmation link. Return to Gmail settings and refresh the page. Select **Forward a copy of incoming mail to**, choose the verified address, keep Gmail's copy in the Inbox and select **Save Changes**.[^6] 4. Review the new mailbox's autonomy setting in Command Mail. New mailboxes start on **auto**. For a reading-only trial, set it to **read**. Autonomy governs organizational operations, not sending; scopes are account-wide rather than isolated permissions for each mailbox.[^7] 5. In ChatGPT web, go to **Plugins → + → Add custom MCP server**. Name it Command Mail. Under **Connection**, enter `https://commandmail.app/mcp`, select OAuth authentication, review the risk warning, tick **I understand and want to continue**, then choose **Create as a plugin**.[^4] 6. Complete the Command Mail sign-in and permission review when prompted by OAuth. Keep only **Read and organize your mail** (`mail:read`) for the first reading task. Switch off **Create and revise drafts**, **Send approved drafts** and **Set up mailboxes** if requested; these start selected. **Send directly, without your approval** (`mail:send`) is off by default and should stay off. Review the discovered tools and install the resulting plugin.[^4][^7] 7. Set the Command Mail connection in ChatGPT to **Allow read actions** if available, so changes require a prompt. Mailbox autonomy **read** and disabled send permissions do not form a complete write lock: `mail:read` still includes account-wide label changes, for example. If the approval control you need is unavailable, keep the task explicitly limited to reading and do not treat the connection as technically write-protected.[^3][^7] 8. Select Command Mail with **@** in a new conversation and ask the agent to call `capabilities` first and report the owner, granted scopes and available mailboxes. Then search for a known message received after forwarding was enabled.[^7] Gmail forwards new messages except spam. Turning it on does not copy your existing mailbox history.[^6] If you need older messages, Command Mail supports a separate MBOX or EML import. Treat that as an additional task, with the destination mailbox chosen deliberately. Imported and forwarded messages remain copies, with no reverse synchronization to Gmail.[^7] ## Did it work? Verify account, message and permissions Check a result against the original email before trusting a larger summary. Compare the sender, date, subject and the part of the message used to answer your question. Ask ChatGPT which message supports each claimed next step. A convincing answer without an identifiable source does not demonstrate that the correct mailbox was queried. For Command Mail, also check that the forwarded message arrived under the intended address. A missing historical thread may simply never have been imported. Keep the first verification read-only: there is no need to send a test reply, archive real work or delete anything to establish that search works. Forwarding configures receiving only; it does not authorize Command Mail to send from your Gmail address. Sending needs a separately configured, send-enabled address, such as one on your own verified domain. Select that sender deliberately, since it may differ from the Gmail address that received the message.[^7] If you later want Command Mail to send a reviewed draft from that enabled address, `mail:send-approved` sends the particular revision approved by a person. The executing service checks both revision and approval state. The broader `mail:send` allows sending without that individual draft approval. Do not grant it to an agent whose job is only to organize mail, and do not assume a mailbox set to **read** blocks sending.[^7][^8] ## Troubleshooting **Gmail is missing or says Disabled by admin.** Check the selected workspace and the app's availability for your account. A ChatGPT administrator can review workspace enablement, while a Google Workspace administrator may need to approve provider permissions. Enabling a plugin and approving Google access are separate steps.[^1][^2] **The connection uses the wrong Google account.** Review the provider address under **Settings → Plugins**. If **Connect another account** is offered, use it and select the intended account. Confirm the resulting connection after returning to ChatGPT. Some apps support multiple accounts, but OpenAI does not make that a universal promise for every Gmail setup.[^2] **Google says all requested permissions are required.** For **All permissions are required**, OpenAI recommends **Reconnect**. For **ChatGPT needs all requested permissions**, choose **Try again**, then **Select all** on Google’s authorization screen to approve every requested permission for the selected apps. If you only want Gmail, reduce the app selection first. If the access is unacceptable, stop rather than granting broader access merely to remove an error.[^2] **Command Mail can search but cannot write.** Start with the ChatGPT plan: custom MCP on Pro is read/fetch only, while full support is available on Business, Enterprise and Edu. Then inspect Command Mail's scopes, plan and mailbox settings. After a server changes its tool definitions, a workspace administrator may need to refresh and approve the frozen tool snapshot.[^4][^7] ## Data controls and disconnecting later Google data has a specific policy. OpenAI says it does not train generalized models on information accessed directly from connected Google apps or derivations of that data, with exceptions for feedback, manually copied, pasted or uploaded data, and material included in ChatGPT responses. Turning off **Improve the model for everyone** prevents app data in conversations being used to improve models.[^5] Do not automatically apply that Google-specific protection to copies forwarded through Command Mail. Review the relevant service policies and ChatGPT data controls for the actual route you choose. Memory controls are also separate from approval settings.[^3][^5] To stop direct Gmail access, open **Settings → Plugins → Gmail**, select the relevant account under **Connected accounts** where available, and use **••• → Disconnect**. Otherwise use the displayed Connection section. Disconnecting stops future access through that connection; existing conversations and saved memories are not automatically removed.[^2] ## Next steps Once a small search returns the right evidence, try a bounded review such as “List unanswered questions from messages received this week in this account.” Specify the account and period so the agent does not have to guess the scope. Expand permissions only when the workflow actually requires a new action. For the forwarded-copy route, continue with [permissions and autonomy](https://commandmail.app/en/docs/permissions-and-autonomy) and [using multiple email accounts with one agent](https://commandmail.app/en/guides/multiple-email-accounts-ai-agent). Both help distinguish the agent's permissions from the mailbox it works on and the original provider account. [Start Command Mail free](https://commandmail.app/login?signup=1) and connect your agent when you are ready. ## Frequently asked questions ### Do I need Developer Mode to connect Gmail? No. Use the Gmail app available under Plugins. The current custom MCP setup also does not require enabling Developer Mode first, although custom servers have separate plan and workspace requirements. ### Does the Command Mail connection work on ChatGPT Plus? OpenAI does not document Plus support for the custom MCP route described here. It lists Pro for read/fetch access and Business, Enterprise and Edu for full MCP support. Check this before setting up forwarding. ### Can I add a second Gmail account? Some apps support multiple connected accounts, depending on the app and your account. Check whether Gmail shows Connected accounts and Connect another account. Do not assume that support is identical in every account or interface. ### Will Command Mail changes sync back to Gmail? No. Command Mail receives forwarded or imported copies. Archiving a message there does not archive the original in Gmail. ### Does disconnecting Gmail remove previous conversations? No. It stops future access through that connection. Existing chats and saved memories are separate and are not automatically deleted. ## Sources 1. [OpenAI: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 2. [OpenAI: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 3. [OpenAI: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 4. [OpenAI: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 5. [OpenAI: Google app data controls FAQ](https://help.openai.com/en/articles/10408842-google-app-data-controls-faq), accessed 2026-10-17 6. [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 7. [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 8. [Command Mail: Plans, agent connections and limits](https://commandmail.app/llms.txt), accessed 2026-10-17 [^1]: [OpenAI: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 [^2]: [OpenAI: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 [^3]: [OpenAI: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 [^4]: [OpenAI: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 [^5]: [OpenAI: Google app data controls FAQ](https://help.openai.com/en/articles/10408842-google-app-data-controls-faq), accessed 2026-10-17 [^6]: [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 [^7]: [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 [^8]: [Command Mail: Plans, agent connections and limits](https://commandmail.app/llms.txt), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Connect Gmail to Claude: built-in connector or Command Mail > How to connect Gmail to Claude: the built-in connector for one account in minutes, or Command Mail for several addresses and set permissions. Steps and fixes. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/guides/connect-gmail-to-claude · Deutsch: https://commandmail.app/de/anleitungen/gmail-mit-claude-verbinden.md Claude has a built-in Gmail connector, and for most people with a single Gmail account it is the right choice: go to Customize → Connectors, pick Gmail, sign in with Google, and Claude can search, draft and, after asking you, send. Choose Command Mail instead when Claude should work across several addresses, including ones outside Google, or when you want to set separately what the agent may change in each mailbox and whether it may send. Command Mail is our product. Both routes are described in full below, including their limits. ## The result With the built-in connector, Claude works directly in your Gmail account. It can search and read mail in plain language, format drafts, manage labels and threads, list saved drafts, and send, reply and forward once you approve.[^1] With Command Mail, Claude works in a separate inbox that receives a copy of your Gmail through forwarding and can hold your other addresses too. You approve scopes when you connect, and set an autonomy level for each mailbox.[^8] ## How the two routes compare If you have one Gmail account and the default approval prompts are enough, use the built-in connector. If you need several addresses or separate rules for sending, use Command Mail. | | Claude's Gmail connector | Command Mail | |---|---|---| | Accounts | the Google account you connected; check multi-account support in your own Claude app | several addresses from different providers, up to ten on Pro | | Setup | a few clicks and a Google sign-in | account, Gmail forwarding, connector | | What Claude works on | your Gmail mailbox itself | forwarded copies, nothing synced back | | Sending | asks before each send by default | not on Free; from Pro, approval of the reviewed revision or independent sending with a daily limit | | Control | approval prompts; tool permissions set by owners on Team and Enterprise | scopes at connection, autonomy per mailbox, activity log; undo for supported sorting actions | | Price | available to all users, per Anthropic | Free; Pro €19 (first 100 paying accounts) or €29 per month | Sources for the table: Anthropic's help pages[^1][^2] and Command Mail's published details.[^8][^9] ## Prerequisites For the built-in connector you need a Claude account on any plan and a Google account with Gmail. On Team and Enterprise, an Owner or Primary Owner enables the Google connectors for the organization before anyone can sign in.[^1] If your Gmail is a Google Workspace account, your Google admin may also have to trust Claude; the fix is under Troubleshooting. For Command Mail you also need a Command Mail account and access to your Gmail settings. Free covers one address and the last 90 days, without sending. Pro adds up to ten mailboxes, full history and sending; on 2026-10-17 it costs €19 per month for the first 100 paying Pro accounts and €29 after that, with a 7-day trial for €1.[^8] ## Route 1: Steps for the built-in Gmail connector 1. Open Claude on the web or Claude Desktop. 2. Go to Customize → Connectors and click the + button next to Connectors.[^2] 3. Find Gmail in the list and review its description and capabilities. 4. Click Connect and sign in to the Google account whose mail Claude should use.[^2] 5. Google's consent screen mentions permission to send email. That is expected: Claude can send, reply and forward, and asks for your approval first by default.[^1] Allow access. 6. Turn Gmail on in a chat: click + in the lower left (or type /), hover over Connectors, and switch Gmail on.[^2] 7. Ask something you can verify, such as "Which emails from this week still need a reply?" Claude cites the emails it used, with links to the originals when available.[^1] On Team and Enterprise plans, the owner adds the connector in Organization settings → Connectors → Browse connectors → Add to your team, and each member then signs in individually.[^2] Owners can also decide, organization-wide, which tools are always allowed, need approval or are blocked, for example allowing search and summaries while blocking sending.[^2] ## What the Gmail connector does well, and its limits For one account, the connector covers daily email work: natural-language search, reading, drafting with context, sending with approval, and tidying labels and threads.[^1] Anthropic lists these current limitations: attachment content is not accessible (metadata only), some advanced Gmail filters may not be supported, complex queries may need several API calls, Google's API quotas apply, and performance may vary for large mailboxes.[^1] Claude can only access the Google account you connected; Anthropic does not state a maximum number of accounts.[^1] An open user issue from March 2026 reports that connecting a second Gmail account requires disconnecting the first.[^5] Check whether your Claude interface offers another account. Another open issue, filed in October 2026, asks for a way to save a forward as a draft, because the connector's forward tool sends rather than preparing a draft for review.[^6] On data handling, Anthropic says data retrieved through connectors is stored with the chat on its servers and is deleted when you delete the chat. It does not train its models on Gmail connector data. On consumer plans, if you allow training on your chats, content you copy and paste from Gmail may be used, and so may copied Claude responses that include specific information from the connector. The data the connector retrieves directly is not used for training.[^1] ## Route 2: Steps for Gmail through Command Mail Here Claude never touches your Google account. Gmail forwards new mail to a private intake address, and Claude works in Command Mail. 1. **Sign up.** On commandmail.app choose Start free. Google sign-in shares only your name, email address, profile picture link and account ID, and gives no Gmail access.[^9] 2. **Add your Gmail address as a mailbox.** In Connect a mailbox, enter your Gmail address. Command Mail shows an intake address that belongs to this mailbox only. Copy it.[^9] 3. **Add the intake address in Gmail.** Settings (top right) → See all settings → Forwarding and POP/IMAP (or just Forwarding) → Add a forwarding address. Paste the intake address and click Next → Proceed → OK.[^7] 4. **Verify.** Gmail sends a verification link to the intake address. Open that message in Command Mail and click the link.[^7] 5. **Turn forwarding on.** Refresh Gmail's settings, select Forward a copy of incoming mail to, choose Keep Gmail's copy in the Inbox, and click Save Changes.[^7] Your Gmail stays complete. 6. **Connect Claude.** Open Command Mail in the Claude directory and choose Connect.[^4] The listing carries the Community label, which means it passed automated checks but is not verified by Anthropic.[^4] You can also add it as a custom connector: Customize → Connectors → Add → Custom → Web, enter a name such as Command Mail and the URL https://commandmail.app/mcp, then Continue. Review the detected authentication settings and click Continue again. Under Authentication choose Sign in now, and keep the recommended OAuth client option, Use Claude's published identity; no API key or request header is needed. Click Add, then sign in to Command Mail. On Claude's free plan you can add only one custom connector.[^2][^3] 7. **Approve permissions.** On Command Mail's authorization page, start with mail:read and mail:draft. Grant sending later and on purpose: mail:send-approved sends exactly the revision you approved, while mail:send lets the agent send on its own, with a unique send key per email and a daily limit of 50 by default.[^9] 8. **Set autonomy.** Under Automation & rules → How much may your agent do?, a new mailbox starts at auto. The agent then carries out sorting actions without asking each time whenever it works; scheduled runs are set up in the agent itself. Archiving, spam, trash, category changes and filing into folders can be undone from the activity log; label changes and sent emails cannot.[^10] suggest is the more careful start: supported sorting actions such as archiving, spam and trash wait as suggestions for your decision.[^9] This level does not limit sending, and with mail:read and mail:send together the REST API currently allows permanent deletion even under suggest. Do not give an agent that should only sort mail:send. Command Mail treats email content as untrusted data and flags suspected prompt injection. When an agent replies to a message recognized as automated, for example by Auto-Submitted or mailing-list headers, Command Mail blocks the send.[^8] This is not a general block on every message sent to automated addresses. ## Check that it worked **Route 1:** Ask Claude to find an email you know exists, for example "Find the last invoice from my phone provider." If it comes back with a source link, the connector works. Ask for a reply draft and then to send it; Claude should ask for your approval first.[^1] **Route 2:** Send yourself a test message from another account. A few minutes later it should be in Command Mail as well. Then ask Claude: "What can you do in Command Mail, and which mailboxes do you see?" It calls capabilities first and lists your scopes and each mailbox with its autonomy level.[^10] ## Troubleshooting **"Please try again. You may need to reconnect with your Google Drive account."** Go to Customize → Connectors, open the Google connector, click Disconnect and confirm. The next time you use it, Claude sends you to Google to sign in again. If that fails, remove the connection at myaccount.google.com/connections with Delete all connections.[^1] **"Access blocked: your institution's admin needs to review Claude for Google Drive."** A Google Workspace admin has to trust Claude: admin.google.com → Security → Access and data control → API controls → Manage third-party app access → Add app → OAuth App Name, search for Claude and set it to Trusted. The change takes about 15 minutes.[^1] **Claude answers without looking at your mail.** The connector is off for this chat. Click + → Connectors and switch it on.[^2] **The Gmail verification message is not in Command Mail.** Check that you pasted the whole intake address and that the mailbox is connected. **Spam and older mail are missing in Command Mail.** Gmail forwards all new messages except spam.[^7] Older mail comes over once from an MBOX export, such as one from Google Takeout, with commandmail import; it is a one-time copy with no ongoing sync.[^9] **Archiving in Command Mail did not change Gmail.** That is how it works: Command Mail holds copies and does not write back to Gmail.[^9] ## When Command Mail is the wrong choice If you have one Gmail account and want Claude to archive and label inside Gmail itself, the built-in connector is shorter and changes your real mailbox. If you do not want copies of your mail with another provider, note that Command Mail is operated by FYN Labs LLC in the United States, with hosting and database at Cloudflare and attachments stored in Cloudflare's EU region.[^11] And if Claude should send mail for you, Command Mail's free plan will not do it. ## Next steps If you also use Outlook or your own domain, [Use multiple email accounts with Claude or ChatGPT](https://commandmail.app/en/guides/multiple-email-accounts-ai-agent) shows how to bring every address together. Commands for Claude Code, Codex, Cursor and other agents are on [For agents](https://commandmail.app/en/agents); scopes, autonomy and the CLI are in the [documentation](https://commandmail.app/en/docs). Start Command Mail free with your Gmail address, then connect it to Claude. ## Frequently asked questions ### Is the Gmail connector free in Claude? Anthropic lists no extra charge. Its help page says the Google Workspace connectors for Gmail, Calendar and Drive are available for all users on Claude and Claude Desktop. On Team and Enterprise, an owner has to enable them for the organization first. ### Will Claude send an email without asking me? Not by default. Claude asks for approval before each send, reply or forward. On Team and Enterprise plans, owners decide whether members may let these actions run without asking. ### Can Claude read Gmail attachments? Not through the Gmail connector. Anthropic lists attachment content as a current limitation: Claude sees attachment metadata only. ### Can I connect two Gmail accounts to Claude? Anthropic documents access to the one Google account you connected and no multi-account option. An open user issue from March 2026 reports that connecting a second account means disconnecting the first. Check whether your Claude interface offers another account; otherwise forward both accounts into one mailbox if Claude should see them together. ### Does Command Mail get access to my Google account? No. It only receives the messages Gmail forwards to your private intake address. Signing in with Google shares your name, email address, profile picture link and account ID, nothing from Gmail. ## Sources 1. [Anthropic: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 2. [Anthropic: Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities), accessed 2026-10-17 3. [Anthropic: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 4. [Claude directory: Command Mail](https://claude.ai/directory/command-mail), accessed 2026-10-17 5. [GitHub issue anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 6. [GitHub issue anthropics/claude-code #100177: Gmail connector: let forward create a draft instead of sending](https://github.com/anthropics/claude-code/issues/100177), accessed 2026-10-17 7. [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 8. [Command Mail: llms.txt with plans, scopes and connection options](https://commandmail.app/llms.txt), accessed 2026-10-17 9. [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 10. [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 11. [Command Mail: Privacy policy](https://commandmail.app/privacy), accessed 2026-10-17 [^1]: [Anthropic: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 [^2]: [Anthropic: Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities), accessed 2026-10-17 [^3]: [Anthropic: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 [^4]: [Claude directory: Command Mail](https://claude.ai/directory/command-mail), accessed 2026-10-17 [^5]: [GitHub issue anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 [^6]: [GitHub issue anthropics/claude-code #100177: Gmail connector: let forward create a draft instead of sending](https://github.com/anthropics/claude-code/issues/100177), accessed 2026-10-17 [^7]: [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 [^8]: [Command Mail: llms.txt with plans, scopes and connection options](https://commandmail.app/llms.txt), accessed 2026-10-17 [^9]: [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 [^10]: [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 [^11]: [Command Mail: Privacy policy](https://commandmail.app/privacy), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Permissions and autonomy for agents in Command Mail > Every Command Mail permission and autonomy level: what an agent may read, sort, draft and send, how approvals and limits work, and how to revoke access. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/docs/permissions-and-autonomy · Deutsch: https://commandmail.app/de/docs/berechtigungen-und-autonomie.md An agent in Command Mail is limited in two ways. **Permissions** (scopes) are chosen when you connect it and apply to every mailbox in your account. **Autonomy levels** are set per mailbox and decide whether certain organizing actions by an agent take effect there, wait as a suggestion or are refused. Autonomy does not limit reading or sending. This page describes how Command Mail actually behaves, including the places where it protects less than a level called "read" might suggest. It expands on the Scopes and Autonomy sections of the [documentation](https://commandmail.app/en/docs) and the [page for agents](https://commandmail.app/en/agents). ## Recommendation: permissions for a sorting agent An agent that only sorts your mail gets **`mail:read`**, plus **`mail:draft`** if it should prepare replies. Do **not** grant it **`mail:send`**. With `mail:read` and `mail:send` together, an agent can delete mail for good and empty Trash or Spam, and on the `suggest` and `rules` levels those actions do not wait for your approval. Keep mailboxes on `suggest` while you want to see every cleanup, and switch to `auto` only when the agent should clean up on its own. If the agent should send replies after you approve them, add `mail:send-approved`. Reserve `mail:send` for an agent that really should send without asking, ideally in a separate Command Mail account. ## Permissions at a glance Every connection carries a list of permissions, whether it uses the hosted MCP server, the CLI or an API key.[^1] A call without the needed permission is refused, and the MCP server does not even offer many tools to an agent that lacks the matching permission. | Permission | Allows | Does not allow | |---|---|---| | `mail:read` | read and search every mailbox; overview and morning brief; organize within the autonomy level; labels and folders; read and star flags; suggest rules; view activity and undo changes made by agents and rules; follow-ups | sending, moving to Trash, deleting for good | | `mail:draft` | save drafts and discard untouched agent drafts in this account | sending, overwriting your edits | | `mail:send-approved` | send the exact revision you approved, once | sending anything you did not approve | | `mail:send` | send new mail and replies without approval from any send-enabled mailbox in the account, each with its own send key, within the daily limit | more than 20 recipients per message, replies to mail recognized as automated | | `mailboxes` | connect addresses, prove domain ownership with a DNS TXT record, import old mail, change signatures | reading or sending mail | Operator permissions such as `admin` are not available for customer keys.[^1] By default, reading through MCP and the CLI leaves the read status alone; the CLI and local MCP offer an explicit option to mark mail as read. A direct REST call to `GET /api/mail/{id}` marks the conversation as read unless you add `markRead=false`; in mailboxes on `read` the flag stays unchanged either way. ### Actions that need two permissions Some actions hand content to new recipients or remove mail. They need `mail:read` and `mail:send` together: forwarding, moving to Trash and restoring, deleting for good, emptying Trash or Spam, marking everything read, and switching on or shortening automatic removal of Spam and Trash.[^3] A key that may only send therefore cannot read your mail or wipe your mailbox. ### Actions limited to the browser session An agent's access cannot approve drafts, confirm or dismiss suggestions, change autonomy levels, set the daily send limit, manage API keys or generate text through a model route.[^2] The server refuses these calls from a token; they only work in a signed-in browser session. ### Choosing permissions when you connect The authorization page lists each requested permission separately, and you can switch off any of them. `mail:send` starts switched off, so by default an agent drafts and you approve.[^3] Permissions apply to every mailbox in your account. ## Autonomy levels per mailbox The autonomy level decides whether an agent's logged organizing actions in a mailbox take effect, wait as a suggestion or are refused.[^1] These actions are archive, spam, Trash, category changes and filing into folders. The order is `read` → `suggest` → `rules` → `auto`. | Level | Agent's logged organizing actions | Read flags, stars and labels | Your rules | |---|---|---|---| | `read` | refused | refused | do not act | | `suggest` | stored as suggestions (`{"suggested": true}`) | take effect | stay suggestions | | `rules` | stored as suggestions | take effect | enabled rules act | | `auto` | run and are logged, at most 50 conversations per organizing call | take effect | act | **Deleting is the exception.** Deleting for good and emptying Trash or Spam bypass the suggestion step. If an agent holds `mail:read` and `mail:send`, Command Mail runs these actions right away on `suggest`, `rules` and `auto`; only mailboxes on `read` are left untouched. Mail deleted for good cannot be recovered. That is why a sorting agent should not get `mail:send`. **What autonomy does not limit.** Autonomy restricts neither reading nor sending. An agent with `mail:read` also reads mailboxes on `read`, and an agent with `mail:send` can send from any send-enabled mailbox in the account and reply to messages, including ones in a mailbox on `read`. If a mailbox must be off limits to an agent, it needs its own Command Mail account. A mailbox you have not configured runs on `auto`, so check the level in Settings after connecting an address. When one organizing call touches mail from several mailboxes, the strictest of their levels applies. Filing into folders takes effect only on `auto`, waits as a suggestion on `suggest` and `rules`, and is refused on `read`. Closing or moving follow-up reminders needs `auto`. An agent may set automatic removal of Spam and Trash only when the account's default level is `rules` or `auto`. An agent should create rules only on your explicit instruction; that is what the guidance Command Mail hands every agent through `capabilities` says, and the server cannot verify it. In mailboxes on `read` or `suggest`, an agent's rule stays a suggestion until you switch it on; on `rules` and `auto`, the agent may switch it on itself. Mail that Command Mail recognized as an invoice, payment problem, security notice, deadline, appointment or personal message is never moved to Spam or Trash by a rule. Archiving or filing such mail needs a rule that names exactly that kind, and no other protection reason may apply; other matches become suggestions.[^2] Recognition works with patterns and will not catch every invoice or deadline. ## Approving drafts: exactly this revision With `mail:draft`, an agent saves a draft. With `mail:send-approved`, it sends that draft after you approve it, exactly the approved revision and only once.[^1] 1. The agent saves a draft, as a reply in the conversation if you like and with up to five attachments. Every change creates a new revision. 2. You review the draft in the app and approve that revision. If a newer message arrived in the conversation after the draft was written, Command Mail points it out and you approve again on purpose. 3. The agent calls `send_approved_draft` with the approved revision. If the draft changed since, or the sending address is no longer valid, Command Mail refuses. 4. If it is unclear whether the message was delivered, Command Mail does not retry. If delivery was refused before sending, the approval stays in place. Once you edit or approve a draft, no agent can overwrite or discard it. Untouched agent drafts can be discarded by any agent in this account that holds `mail:draft`, not only the one that wrote them. Approved mail goes out without a signature added afterwards, exactly as you reviewed it, and it does not count toward the agent daily limit. Sending requires a Pro plan and a sending address that is set up for it.[^3] ## Sending on its own: send key, daily limit, recipient cap `mail:send` is meant for agents that should reply without asking, such as in customer support.[^1] Three brakes apply. **Send key.** Every message an agent sends needs its own `sendKey` of 8 to 128 letters, digits or the characters `_ . : -`. If an agent repeats a request with the same key, for example after a lost response or from a second window, the message is not sent twice. When delivery was certainly refused, Command Mail releases the key right away. Keys not bound to a stored message are cleaned up after 30 days. After an unclear outcome, an agent should check the send status and must not start a new request with a new key. **Daily limit.** By default an agent sends at most 50 messages per day, counted in UTC. You set the value between 0 and 500 in the automation settings; 0 stops all agent sending.[^1] Messages with an unclear outcome count. Your own messages and approved drafts do not. The agent sees the current numbers in `capabilities` under `limits`. **Recipient cap.** A message an agent sends on its own may have at most 20 recipients across To, Cc and Bcc.[^3] Above that, nothing is sent, and the agent should save a draft for you. ## Machine-sent mail Command Mail blocks direct agent replies to messages it recognized as machine-sent from their headers.[^2] It looks for `Auto-Submitted` under RFC 3834 with any value other than `no`, `Precedence: bulk`, `junk` or `list`, `List-Id` and common autoresponder markers. When one of these is present and an agent with `mail:send` replies to that message, nothing is sent; the agent can still save a draft for you to approve. This check has limits. Without those headers, Command Mail does not recognize an out-of-office note or newsletter as automated. It also covers replies only: a new message to the same address is not checked against it. It is not complete protection against mail loops. ## Prompt injection: email content is data Email content cannot, through Command Mail, grant an agent permissions, approve drafts, create rules or change model routes.[^3] Command Mail scans subject and text for patterns aimed at an AI model, such as "ignore all previous instructions", requests to forward every invoice, or requests for passwords and codes. Such messages get a visible flag, `get_thread` reports it to the agent, and the daily overview lists the attempts. The flag is a warning. It does not catch every phrasing, and an agent that is swayed by a message can still act within the permissions you granted. Narrow permissions therefore remain the real limit. The risk assessment looks at text, link addresses and attachment names, but never follows links or opens attachments.[^1] ## Activity log and undo Archive, spam, Trash, category changes and folder filing are logged with the state before and after.[^1] Undo restores the earlier state only where nothing else changed it since; conflicts are counted. Undoing a change a rule made pauses that rule until you review it. A token may undo what agents, rules and automatic classification did; your own decisions you undo yourself in the app. Suggestions wait in the log for your decision. Sent mail, mail deleted for good, emptied Trash or Spam, label changes and creating or renaming folders cannot be undone. Messages an agent sent on its own are in the Sent folder; approved drafts also appear in the log. ## Common setups **Read and draft only.** Grant `mail:read` and `mail:draft`, plus `mail:send-approved` if you want the agent to send what you approve. Set mailboxes to `suggest`. The agent reads, proposes cleanups and writes replies for your approval. No message leaves without your approval, and without `mail:send` the agent cannot delete anything for good. It still sets read flags, stars and labels directly on `suggest`. **Archive newsletters automatically.** Grant `mail:read` and no `mail:send`. Tell the agent which senders to archive, for example "From now on, archive the newsletters from this shop." On `rules`, it creates a rule and switches it on, so future mail from that sender goes straight to the archive, and it can apply the rule to existing mail in one step you can undo. On `suggest`, the rule waits for you to switch it on. **Forward invoices.** Forwarding needs `mail:read` and `mail:send` together, a Pro plan and a send-enabled address, and each forward counts toward the daily limit. Keep in mind that the same permissions allow permanent deletion. Rules cannot forward; a scheduled run in your agent does that. The safer variant uses approval: grant `mail:read`, `mail:draft` and `mail:send-approved`, let the agent save a new draft with the invoice attached, and approve it. **Let a support mailbox answer on its own.** Grant `mail:read`, `mail:draft` and `mail:send` and choose a daily limit that fits your volume. Because `mail:send` applies to every mailbox in the account and, together with `mail:read`, allows deletion, keep the support mailbox in its own Command Mail account, separate from your personal mail. ## Revoking access In Settings, disconnect an app under connected apps. This revokes every token that app holds for your account, including a sign-in that has not finished yet, and the app must ask for access again. Revoke API keys in the API keys section of Settings; a key cannot manage other keys. To give an agent fewer permissions, disconnect it, connect it again and pick only what it needs on the authorization page. Setup steps are in the [documentation](https://commandmail.app/en/docs), and agent workflows are on the [page for agents](https://commandmail.app/en/agents). [Start free](https://commandmail.app/login?signup=1) and connect your agent with `mail:read` and `mail:draft` first. ## Frequently asked questions ### Which permissions should an agent get that only sorts mail? Only mail:read, plus mail:draft if it should prepare replies. No mail:send: together with mail:read it also allows permanent deletion, and that does not wait for your approval. Keep mailboxes on suggest while you want to see every cleanup. ### Does the read level fully protect a mailbox? No. read stops an agent from sorting, flagging or deleting mail there. The agent can still read that mailbox, and with mail:send it can send from it. Separate read or send rights for individual mailboxes require separate Command Mail accounts. ### Can an agent approve its own draft? Not through its own access. Approving drafts, deciding suggestions, changing autonomy and setting the daily limit only work in a signed-in browser session. Email content never grants permissions. ### What happens if I edit a draft after approving it? Every edit creates a new revision and clears the approval. Only the revision you approved last can be sent. ### How do I remove an agent's access? In Settings, disconnect the app under connected apps; that revokes all of its access. API keys are revoked in the API keys section. ## Sources 1. [Command Mail: Documentation (scopes, autonomy, let your agent reply)](https://commandmail.app/en/docs), accessed 2026-10-17 2. [Command Mail: For agents (permissions, autonomy, limits)](https://commandmail.app/en/agents), accessed 2026-10-17 3. [Command Mail: llms.txt](https://commandmail.app/llms.txt), accessed 2026-10-17 [^1]: [Command Mail: Documentation (scopes, autonomy, let your agent reply)](https://commandmail.app/en/docs), accessed 2026-10-17 [^2]: [Command Mail: For agents (permissions, autonomy, limits)](https://commandmail.app/en/agents), accessed 2026-10-17 [^3]: [Command Mail: llms.txt](https://commandmail.app/llms.txt), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Superhuman alternatives: Spark, Shortwave, Canary, Fyxer > Looking for a Superhuman alternative? Prices, strengths and limits of Spark, Shortwave, Canary, Fyxer and Command Mail, and who each one suits. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/alternatives/superhuman · Deutsch: https://commandmail.app/de/alternativen/superhuman.md Superhuman is a fast AI email client for Gmail and Outlook accounts; Starter costs $30 billed monthly or $300 a year, and its MCP server for Claude, ChatGPT and other agents needs the $40 Business plan.[^1][^2] If price is the issue, look at Spark or Canary; if your team lives in Google Workspace, Shortwave is the closest fit, and if you mostly want reply drafts and meeting notes inside Gmail or Outlook, try Fyxer. Command Mail fits when your own agent should work across several addresses and you want to set, per permission and per mailbox, what it is allowed to do. Command Mail is our product. We checked every claim about the other products on their pricing and help pages on 2026-10-17 and link the source next to it. ## What Superhuman does well Superhuman is built for people who write a lot of email and want to get through it quickly, such as executives and sales teams. Business adds Auto Drafts, Ask AI for questions about your inbox, Custom Auto Labels, HubSpot and Salesforce integration and a feed of recent opens.[^1] Enterprise adds SSO, advanced security controls and dedicated support.[^1] The Superhuman Mail MCP gives agents real access. Its help center lists setup steps for Claude Code, Claude Desktop, ChatGPT, Codex, Cursor, Gemini CLI, Perplexity and other MCP clients. Once connected, an assistant can search mail, draft replies in your voice, manage calendar events and send messages, including scheduled sends, and you can link several Gmail and Outlook accounts that belong to your own seat.[^2] Drafts created over MCP show up in the Superhuman app, where you can edit and send them. Superhuman also documents the controls. The send tool can send directly, you must approve the tool in your AI client, and Superhuman advises a confirmation step for every call. You can also run the MCP server in read-only mode, and an undo tool cancels a message within the undo window.[^2] Approval of individual sends is configured in Claude or ChatGPT. ## Why people look for Superhuman alternatives Cost is one possible reason. Starter is $30 billed monthly or $300 per year, and Business is $40 per month or $396 per year.[^1] Because the MCP server is not available on Starter, anyone who mainly wants an agent in their inbox pays the Business price.[^2] Account coverage is another. Superhuman works with Gmail and Outlook accounts.[^2] Addresses at iCloud, Fastmail, a regional provider or your own mail server have no direct home there. A third reason concerns agents. With Superhuman, you choose between read-only mode and sending with approval in your AI client.[^2] Some people also want a limit inside the mailbox itself, such as a fixed daily cap on sends without approval, or an approval tied to one exact draft revision. ## Superhuman alternatives compared | Product | Price (checked 2026-10-17) | Accounts | Built-in AI | Agents and integrations | |---|---|---|---|---| | Superhuman Mail | Starter $30/month or $300/year, Business $40/month[^1] | Gmail, Outlook[^2] | Auto Drafts, Ask AI, Auto Labels (Business)[^1] | MCP server from Business, can send directly, read-only mode available[^2] | | Spark Mail | Free €0, Plus €10/month, Pro €22/month (prices shown in Germany)[^3] | unlimited accounts, also on Free[^3] | Spark +AI, AI meeting notes; Autopilot announced[^3] | Spark CLI for agents: read access, triage access from Pro; no MCP server found[^3] | | Shortwave | Business plans: $30, $45, $120 per seat per month[^4] | Gmail and Google Workspace; other providers only by forwarding into Gmail[^5] | AI assistant, AI search, AI filters[^4] | MCP integrations for its built-in assistant (client side); no server for outside agents found[^4] | | Canary Mail | Free $0, Growth $36/year, Pro+ $100/year, lifetime option[^6] | Gmail, Outlook, iCloud, Exchange, IMAP[^7] | Copilot drafts, on-device triage[^7] | no agent access found | | Fyxer AI | Free $0, Pro $37, Team $64 per person per month billed yearly[^8] | Gmail, Outlook[^8] | drafts in your style, meeting notes, follow-ups[^8] | MCP for Claude, ChatGPT and your own tools from Team, API on Enterprise[^8] | | Command Mail | Free €0, Pro €19 or €29/month[^10] | any address by forwarding, up to ten mailboxes on Pro[^10] | morning brief, risk scores, sorting[^12] | hosted MCP server, CLI and REST with permissions per connection and autonomy per mailbox[^11] | "Not found" means we found nothing on the vendor's own pages on 2026-10-17. That may change. ## Spark: many accounts at a low price Spark by Readdle is a traditional email client with AI features. Even the free plan takes unlimited email accounts. In Germany, Plus costs €10 per month or €99 per year and Pro €22 per month or €229 per year; visitors from other regions may see prices in their own currency.[^3] Paid plans include the Spark +AI assistant and AI meeting notes, and the Pro column lists Autopilot and Auto-Drafts as "soon".[^3] For agents, the pricing page lists a Spark CLI with read access, and triage access on Pro.[^3] We found no MCP server that Claude or ChatGPT could add directly. Pick Spark if you want several accounts in one app, built-in AI is enough, and an agent working from the command line should read or triage your mail.[^3] ## Shortwave: built for Gmail teams Shortwave is tied closely to Google. Its Business plans cost $30, $45 (Premier) and $120 (Max) per seat per month; plans for individuals sit on a separate tab of the pricing page, and there is a 14-day trial.[^4] You get an AI assistant that searches your mail, AI filters that act on new mail based on your own prompt, and MCP integrations that let this built-in assistant use other services.[^4] The limit is documented in Shortwave's help: other providers work by forwarding their mail to a Gmail account. Outlook.com and Proton Mail are forwarding-only and cannot send from their address, and Microsoft 365 and Exchange are not supported at this time.[^5] For a team that runs entirely on Google Workspace, Shortwave is a strong choice. ## Canary Mail: low price, privacy focus Canary has the lowest paid tier in this list. Growth costs $36 a year and Pro+ $100 a year, and there is a free plan plus lifetime licenses. According to its FAQ, Canary has no monthly subscriptions; the pricing page also shows monthly reference figures of $3 and $10; the annual charges are $36 and $100.[^6] Canary supports Gmail, Outlook, iCloud, Exchange and IMAP on Mac, iPhone, iPad, Windows and Android.[^7] According to Canary, triage and prioritization run on your device, while drafting and summaries use cloud AI without training general models on your mail.[^7] It is a good fit for a budget client that handles many account types. Canary does not describe a way to connect your own agent. ## Fyxer AI: drafts and meeting notes without a new client Fyxer stays inside Gmail or Outlook and works like an assistant: it sorts mail, drafts replies in your style, takes meeting notes and keeps a list of follow-ups you promised.[^8] Pro is $37 per person per month billed yearly and Team is $64; the free plan runs on a small monthly credit allowance.[^8] Fyxer states SOC 2 Type II and ISO 27001 certification and says your data is never used to train third-party AI models.[^9] From the Team plan, the pricing page also lists MCP for using your Fyxer context in Claude, ChatGPT and your own tools, and Enterprise adds a full API.[^8] The page does not say what an agent may do through it, such as sending or moving mail. For customer-facing teams that do not want to switch clients, Fyxer is a solid package. ## Command Mail: one inbox your agent works in Command Mail is one inbox for all of your addresses, built so that your AI agent can work in it.[^10] You forward Gmail, Outlook, iCloud, a custom domain or any other provider to a private inbound address, then connect Claude, ChatGPT, Codex, Cursor or another agent through the hosted MCP server or the CLI.[^11] The difference from Superhuman is where the limits live. When you connect an agent, you choose individual permissions: read, save drafts, send exactly the revision you approved, or send on its own. Sending on its own starts switched off.[^10] If you grant it, every message needs its own send key and a daily limit applies, 50 messages by default.[^11] Each mailbox also has an autonomy level that sets how far an agent may organize mail there on its own. It does not restrict sending: send permissions apply to the whole account. Mail content is treated as data, suspected prompt injection is flagged, and a direct agent reply to a message that Command Mail recognized as machine-sent from its headers is blocked.[^10] [Permissions and autonomy](https://commandmail.app/en/docs/permissions-and-autonomy) explains every permission and level in detail. You can start free with one mailbox, the last 90 days, search, sorting, the morning brief and drafts; the free plan does not send. Pro costs €19 per month for the first 100 paying Pro accounts for as long as the subscription runs and €29 per month after that, and adds sending, up to ten mailboxes and full history.[^10] In ChatGPT, Command Mail currently needs developer mode and an eligible plan because our directory listing is still under review.[^11] ## When Superhuman is the better choice Stay with Superhuman if you want a polished client for Gmail or Outlook where you do most of the writing yourself, and if your team needs HubSpot or Salesforce inside the inbox.[^1] Superhuman's MCP works on your accounts directly and covers your calendar as well.[^2] Command Mail works on forwarded copies and has no calendar. If you are happy with Superhuman and only want to add agent access, the Mail MCP on Business gets you there without moving anything. In that case, turn on confirmation before sending in your AI client, as Superhuman itself recommends, or use read-only mode.[^2] ## When Command Mail fits Command Mail fits when your agent should do most of the inbox work and you rarely write in a client yourself. A common case is several addresses at different providers, such as a Gmail address, a company domain and an old Yahoo address, sorted by one agent. Another is a morning brief with deadlines, open replies and invoices, or a support mailbox where the agent may send answers.[^12] If your personal mailbox must stay out of reach, put the support mailbox in a separate Command Mail account, because an agent with send permission can send from any send-enabled mailbox in its account. Price helps too if Superhuman's client features matter little to you: Command Mail Pro is €19 or €29 per month, and you can try Pro for seven days for €1.[^10] ## Who Command Mail is not for Command Mail does not replace a full email client with calendar, CRM integration and meeting notes. If you need those, Superhuman, Spark or Fyxer serve you better. Because Command Mail works through forwarding, archiving or deleting there does not change your Gmail or Outlook account, and importing old mail is not an ongoing sync.[^12] If you must show certifications such as SOC 2 or ISO 27001, Fyxer documents them and Command Mail does not.[^9] Command Mail is operated by FYN Labs LLC in the United States, and the app and database run on Cloudflare. ## Moving from Superhuman to Command Mail You can keep Superhuman while you test. Command Mail only receives forwarded copies, so both run side by side.[^12] 1. Create an account on commandmail.app with Google, Apple or email. 2. Connect an address. The app, or an agent with the `mailboxes` permission, creates the mailbox and returns its private inbound address.[^12] 3. Set up forwarding in Gmail or Outlook to that inbound address. 4. Connect your agent: in Claude, add a custom connector with `https://commandmail.app/mcp`; in Claude Code, run `claude mcp add --transport http commandmail .[^11] 5. Choose permissions. Start with reading and drafting and add sending later. 6. Optionally import old mail from MBOX or EML files with the CLI.[^12] You know it works when the first forwarded message appears in your inbox and your agent's first `capabilities` call names your account, its permissions and your mailboxes.[^11] The [documentation](https://commandmail.app/en/docs) and the [page for agents](https://commandmail.app/en/agents) cover the full setup. ## Which alternative should you choose? - **Many accounts, small budget, AI inside the client:** Spark or Canary; Spark also offers a CLI for agents.[^3][^6] - **Everything in Google Workspace, team AI search and filters:** Shortwave.[^4][^5] - **Reply drafts and meeting notes in Gmail or Outlook, certifications required, MCP on Team:** Fyxer.[^8][^9] - **Fast client with CRM and its own MCP, budget available:** stay with Superhuman.[^1][^2] - **Your own agent, several providers, approval bound to one draft revision and a daily cap:** Command Mail.[^11] If you want to try Command Mail, [start free](https://commandmail.app/login?signup=1) and connect your agent afterwards. ## Frequently asked questions ### Does Superhuman have an MCP server? Yes. The Superhuman Mail MCP connects Claude, ChatGPT, Codex, Cursor and other MCP clients to your inbox and calendar. According to Superhuman's help center it requires the Business plan or higher; Starter does not include it. ### What is the cheapest Superhuman alternative? Spark and Canary both have free plans and low paid tiers. Command Mail also starts free, but the free plan does not send email; sending starts with Pro. ### Can I keep Superhuman while I try Command Mail? Yes. Command Mail receives copies of your mail through forwarding, so your Gmail or Outlook account and Superhuman stay as they are. Changes you make in Command Mail do not sync back to the original mailbox. ### Which alternative works with Microsoft 365? Spark, Canary and Fyxer connect to Outlook accounts directly. Shortwave's own help page says Microsoft 365 and Exchange are not supported at this time. Command Mail accepts any address your provider can forward. ### Can an agent send through Command Mail without asking me? Only if you grant it the mail:send permission, which starts switched off when you connect. Without it, the agent saves drafts and Command Mail sends exactly the revision you approved. ## Sources 1. [Superhuman: Pricing Plans (help center)](https://help.superhuman.com/hc/en-us/articles/46178817646861-Pricing-Plans), accessed 2026-10-17 2. [Superhuman: Superhuman Mail MCP Server (help center)](https://help.superhuman.com/hc/en-us/articles/46005696690317-Superhuman-Mail-MCP-Server), accessed 2026-10-17 3. [Spark Mail: Pricing](https://sparkmailapp.com/pricing), accessed 2026-10-17 4. [Shortwave: Pricing](https://www.shortwave.com/pricing/), accessed 2026-10-17 5. [Shortwave: Use Shortwave with other email providers](https://www.shortwave.com/docs/how-tos/microsoft-outlook-exchange-other-sign-in-support/), accessed 2026-10-17 6. [Canary Mail: Pricing](https://canarymail.io/pricing), accessed 2026-10-17 7. [Canary Mail: AI](https://canarymail.io/ai), accessed 2026-10-17 8. [Fyxer AI: Pricing](https://www.fyxer.com/pricing), accessed 2026-10-17 9. [Fyxer AI: Security](https://www.fyxer.com/security), accessed 2026-10-17 10. [Command Mail: llms.txt (pricing, permissions, connection)](https://commandmail.app/llms.txt), accessed 2026-10-17 11. [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 12. [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 [^1]: [Superhuman: Pricing Plans (help center)](https://help.superhuman.com/hc/en-us/articles/46178817646861-Pricing-Plans), accessed 2026-10-17 [^2]: [Superhuman: Superhuman Mail MCP Server (help center)](https://help.superhuman.com/hc/en-us/articles/46005696690317-Superhuman-Mail-MCP-Server), accessed 2026-10-17 [^3]: [Spark Mail: Pricing](https://sparkmailapp.com/pricing), accessed 2026-10-17 [^4]: [Shortwave: Pricing](https://www.shortwave.com/pricing/), accessed 2026-10-17 [^5]: [Shortwave: Use Shortwave with other email providers](https://www.shortwave.com/docs/how-tos/microsoft-outlook-exchange-other-sign-in-support/), accessed 2026-10-17 [^6]: [Canary Mail: Pricing](https://canarymail.io/pricing), accessed 2026-10-17 [^7]: [Canary Mail: AI](https://canarymail.io/ai), accessed 2026-10-17 [^8]: [Fyxer AI: Pricing](https://www.fyxer.com/pricing), accessed 2026-10-17 [^9]: [Fyxer AI: Security](https://www.fyxer.com/security), accessed 2026-10-17 [^10]: [Command Mail: llms.txt (pricing, permissions, connection)](https://commandmail.app/llms.txt), accessed 2026-10-17 [^11]: [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 [^12]: [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Use multiple email accounts with Claude or ChatGPT > Claude's Gmail connector reaches the account you connected, ChatGPT sometimes more. How to give your AI agent every address, with options, limits and steps. Updated: 2026-10-17 · Fact-checked: 2026-10-17 · Canonical: https://commandmail.app/en/guides/multiple-email-accounts-ai-agent · Deutsch: https://commandmail.app/de/anleitungen/mehrere-e-mail-konten-ki-agent.md Claude's built-in Gmail connector works with the Google account you connected, and users report that a second account only works after disconnecting the first. ChatGPT handles several Gmail accounts only where its Gmail app offers that for your account. Addresses at other providers, such as a domain hosted by your web host, are not part of either connector. To give an AI agent all your mail, you first bring the addresses together in one place, and then connect the agent to that place. Command Mail is our product. This guide compares the options, their limits and the setup steps, then walks through the setup with Command Mail. ## What you will have at the end One connector in Claude or ChatGPT gives the agent every address you own: say a personal Gmail address, a Microsoft 365 work address and billing@ on your own domain. You can ask "What arrived across all my inboxes since yesterday?" or "Only show mail sent to my work address", and the agent answers from a single source. You also decide, mailbox by mailbox, how much the agent may change on its own, from read-only to sorting by itself. Permission to send is granted separately. ## Where native connectors fall short Anthropic states that Claude can only access the Gmail, Calendar and Drive data of the Google account you connected; the help page does not state a maximum number of accounts.[^1] An open user issue in the Claude Code repository, filed in March 2026, reports that only one Gmail account can be connected and that connecting a second requires disconnecting the first. The request is to connect several accounts and target each one by address.[^4] Check whether your Claude interface now offers another account. Adding your own connector as a workaround is tight on the free plan, because Anthropic limits Free users to one custom connector.[^2] ChatGPT is less clear cut. OpenAI writes that some apps support multiple connected accounts, depending on the app and your account; where this is available, the app's page under Settings → Plugins shows Connected accounts and an option to connect another account.[^6] An issue in the Codex repository from June 2026 reports the opposite experience with Gmail, a single active account, and asks for named accounts so replies come from the address that received the mail.[^8] Check what your own ChatGPT offers. Even with two Gmail accounts connected, Outlook, your own domain and other providers remain separate apps or are not available at all. ChatGPT lets you choose when it asks before an action, from Always ask to Allow all actions.[^5] Where the app supports it, you can set this per connected account, so a second account in the same app can have a different setting.[^15] Command Mail adds its own autonomy level per mailbox, which applies the same way to every agent you connect. ## Four ways to give an agent several addresses The options differ mainly in who holds your mail and how precisely you control what the agent does. | Option | Different providers | Effort | Main limitation | |---|---|---|---| | Switch the connected account | no | low | one account at a time | | Forward everything into one Gmail account | yes | medium | all mail sits together in the chosen Google account | | Run your own MCP server | depends on the project | high | you host it and register your own OAuth client | | Use Command Mail as a shared inbox | yes | medium | works on forwarded copies, no sync back | **Switch accounts.** Disconnect one Gmail account and connect the other when you need it. It is free and fine if the second account matters rarely. A question across all inboxes is not possible this way. **Forward into one Gmail account.** Send your other addresses to a main Gmail account and connect only that one to Claude or ChatGPT. This is the least expensive route to a combined view. The connector's own limits stay in place; in Claude, for example, attachments show metadata only.[^1] All forwarded mail then sits together in the Google account you chose. For work mail, your organization's rules have to allow that. **Run your own MCP server.** Open source projects such as google_workspace_mcp connect Gmail and other Google services to an agent. According to its README you create your own Google Cloud project and OAuth client and host the server yourself.[^11] That suits developers who want full control and accept the maintenance. **Use an inbox built for agents.** Command Mail gives each address its own private intake, receives mail by forwarding, and exposes all mailboxes through one connector.[^13] You approve scopes when you connect, and you set an autonomy level per mailbox.[^12] The trade-off is that Command Mail works with copies. Archiving or deleting there leaves the original mailbox as it was. ## What you need - A Command Mail account. The free plan includes one mailbox, the last 90 days, reading, search, sorting and drafts, with no sending; extra mailboxes pause but keep receiving. Several addresses need Pro, which covers up to ten mailboxes with full history and sending. On 2026-10-17 Pro costs €19 per month for the first 100 paying Pro accounts and €29 after that, and a 7-day trial costs €1.[^12] - Access to the settings of each existing mailbox so you can turn on forwarding. - An agent that supports remote MCP. Claude supports custom and directory connectors on every plan; on Team and Enterprise, an owner adds the connector for the organization first.[^2] In ChatGPT, Command Mail's public directory listing is still under review, so you add it yourself as a custom MCP server. According to OpenAI you no longer need to turn on Developer Mode for this. Full MCP support, including write actions, is currently available to ChatGPT Business, Enterprise and Edu on the web. On Business only admins and owners can add custom MCP servers; on Enterprise and Edu, admins can grant that right to members. Pro users can connect custom servers with read and fetch permissions only.[^7] ## Steps Connect the addresses first, then forward, then add the agent. That way the inbox already holds mail the first time the agent looks. ### Step 1: Create an account 1. Go to commandmail.app and choose Start free. You can sign up with Google, Apple or email.[^12] 2. Signing in with Google shares only your name, email address, profile picture link and account ID. It gives Command Mail no access to Gmail, Drive or Calendar.[^13] ### Step 2: Connect each address 1. During first setup, enter your first address in the Connect a mailbox step, for example your Gmail address. 2. Command Mail returns a private intake address for it. Copy it, then continue to your inbox. 3. For every further address, open Settings → Connections. In the Addresses section, enter the address under Connect a mailbox and click Add. 4. The new address appears in the list with its own intake address. Copy it for forwarding. If you prefer the terminal, the CLI does the same, once per address, and needs Node.js 18 or later:[^13] ~~~sh curl -fsSL https://commandmail.app/install.sh | sh commandmail login commandmail mailboxes connect billing@yourcompany.com ~~~ ### Step 3: Turn on forwarding at each provider **Gmail:** Settings → See all settings → Forwarding and POP/IMAP (or just Forwarding) → Add a forwarding address. Paste the intake address and click Next → Proceed → OK. Gmail sends a verification link to that address; it arrives in Command Mail, where you open it. Back in Gmail, refresh the page, select Forward a copy of incoming mail to, choose Keep Gmail's copy in the Inbox, and click Save Changes.[^9] **Outlook.com:** Settings → Mail → Forwarding and IMAP → Enable forwarding, enter the intake address, Save. In new Outlook and Microsoft 365 the path is Settings → Mail → Forwarding, and you can tick Keep a copy of forwarded messages.[^10] **A domain at your web host:** Add a forward to the intake address in your host's control panel, and keep a copy in the old mailbox if it should stay complete. ### Step 4: Connect your agent **Claude (web and desktop):** Open Command Mail in the Claude directory and choose Connect. Or go to Customize → Connectors → Add → Custom → Web and enter .[^14][^3] **ChatGPT (on the web, with an eligible plan and permission):**[^7] 1. Open Plugins, select the plus button and choose Add custom MCP server. 2. Enter a name such as Command Mail and, optionally, a description. 3. Under Connection, enter https://commandmail.app/mcp and set up OAuth sign-in. 4. Read the risk warning, tick I understand and want to continue, and select Create as a plugin. Review the tools it found. 5. Install the plugin, sign in to Command Mail, and start a new conversation. Type @ and select the plugin. **Claude Code:**[^12] ~~~sh claude mcp add --transport http commandmail https://commandmail.app/mcp ~~~ ### Step 5: Choose permissions After you sign in, Command Mail shows an authorization page. Start with mail:read, which covers reading, searching and sorting within the mailbox's autonomy level, and mail:draft, which saves drafts and never sends. mail:send-approved sends exactly the draft revision you approved. mail:send lets the agent send on its own within a daily limit; leave it off until you have a clear use for it.[^12] ### Step 6: Set autonomy for each mailbox Open Automation & rules and, under How much may your agent do?, pick a level for each mailbox. New mailboxes start at auto: the agent carries out sorting actions without asking each time. Supported sorting actions such as archiving, spam, moving to the trash and filing into folders are logged and can be undone.[^14] Sent emails and permanent deletion cannot be undone, and label changes are not covered by undo either. To watch first, choose suggest: supported sorting actions such as archiving, spam and trash then wait as suggestions. With read, Command Mail refuses these sorting actions from the agent.[^13] Neither level limits sending: an agent with mail:send can send from any configured sender address, even if its mailbox is set to read. An agent holding both mail:read and mail:send can currently also delete mail permanently through the REST API under suggest. Do not give a sorting or reading agent mail:send. ## Check that it worked Send a test message from another account to each address. Within a few minutes each one should appear in the matching mailbox in Command Mail. Then ask the agent: "Which mailboxes can you see in Command Mail, and what are you allowed to do?" It calls capabilities first, and the answer lists the owner, your granted scopes and every mailbox with its autonomy level.[^14] On the free plan, a second mailbox is listed too, marked as paused. A second check: "Summarize what arrived since yesterday, grouped by address." A summary that covers every address means you are done. ## Common problems **Gmail's verification message never shows up.** Make sure the address is connected in Command Mail and that you pasted the right intake address, since Gmail sends the link to exactly that address.[^9] **Spam is missing.** Google forwards all new messages except spam.[^9] **An address is missing from the agent's answer.** First check whether it is listed under Settings → Connections → Addresses. **An address is marked as paused.** Only one mailbox is active on the free plan. Others keep receiving, but their mail stays hidden until you move to Pro.[^12] **Older mail is missing.** Forwarding only covers new mail. Bring older messages over once from an MBOX or EML export, for example with commandmail import export.mbox --mailbox billing@yourcompany.com. An import is not an ongoing sync.[^13] **Nothing changes in the original mailbox.** That is by design: Command Mail works on copies and writes nothing back. **Claude ignores the connector.** Turn it on for the conversation with the plus button → Connectors.[^3] **ChatGPT offers no write actions.** On plans without full MCP support, OpenAI allows read and fetch only for custom servers.[^7] **The agent says it cannot send.** The free plan does not send. On Pro the agent also needs mail:send-approved or mail:send and a sender address you have set up.[^12] ## Next steps If Gmail and Claude are your main pair, [Connect Gmail to Claude](https://commandmail.app/en/guides/connect-gmail-to-claude) compares the built-in connector with Command Mail step by step. Every connection option, scope and limit for agents is on [For agents](https://commandmail.app/en/agents); forwarding, the CLI and autonomy are covered in the [documentation](https://commandmail.app/en/docs). You can start Command Mail free with one address and connect your agent in a few minutes. ## Frequently asked questions ### Can Claude connect to two Gmail accounts at the same time? Anthropic describes access to the Google account you connected and documents no multi-account option. An open user issue from March 2026 reports that connecting a second account means disconnecting the first. Check whether your Claude interface offers another account; if not, combine your addresses in one mailbox first. ### Does ChatGPT support more than one Gmail account? OpenAI says some apps support multiple connected accounts and that this depends on the app and your account. Open Settings → Plugins → Gmail. If you see Connected accounts with an option to connect another account, you can add a second one. ### What is the cheapest way to see all my mail in one place for Claude? Forward your other addresses into one Gmail account and connect only that account to Claude. It costs nothing extra, but business and personal mail then share one Google account, and the connector's limits still apply. ### Does Command Mail change my original Gmail or Outlook mailbox? No. It receives copies through forwarding. Archiving, sorting or deleting in Command Mail does not touch the original mailbox, and importing old mail is a one-time copy with no ongoing sync. ### How many addresses does Command Mail support? Free covers one mailbox and the last 90 days; extra mailboxes pause but keep receiving. Pro covers up to ten mailboxes with full history and sending, at €19 per month for the first 100 paying Pro accounts and €29 after that (checked 2026-10-17). ## Sources 1. [Anthropic: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 2. [Anthropic: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 3. [Anthropic: Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities), accessed 2026-10-17 4. [GitHub issue anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 5. [OpenAI: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 6. [OpenAI: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 7. [OpenAI: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 8. [GitHub issue openai/codex #30418: Gmail connector should support multiple named accounts](https://github.com/openai/codex/issues/30418), accessed 2026-10-17 9. [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 10. [Microsoft: Turn automatic forwarding on or off in Outlook](https://support.microsoft.com/en-us/outlook/mail/turn-automatic-forwarding-on-or-off-in-outlook), accessed 2026-10-17 11. [GitHub: google_workspace_mcp (README)](https://github.com/taylorwilsdon/google_workspace_mcp), accessed 2026-10-17 12. [Command Mail: llms.txt with plans, scopes and connection options](https://commandmail.app/llms.txt), accessed 2026-10-17 13. [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 14. [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 15. [OpenAI: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 [^1]: [Anthropic: Use Google Workspace connectors](https://support.claude.com/en/articles/10166901-use-google-workspace-connectors), accessed 2026-10-17 [^2]: [Anthropic: Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp), accessed 2026-10-17 [^3]: [Anthropic: Use connectors to extend Claude's capabilities](https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities), accessed 2026-10-17 [^4]: [GitHub issue anthropics/claude-code #36024: Support multiple Gmail accounts in MCP integration](https://github.com/anthropics/claude-code/issues/36024), accessed 2026-10-17 [^5]: [OpenAI: Connected apps in ChatGPT](https://help.openai.com/en/articles/11487775-connected-apps-in-chatgpt), accessed 2026-10-17 [^6]: [OpenAI: Connecting and managing app accounts in ChatGPT](https://help.openai.com/en/articles/20001494-connecting-and-managing-app-accounts-in-chatgpt), accessed 2026-10-17 [^7]: [OpenAI: Connecting custom MCP servers in ChatGPT](https://help.openai.com/en/articles/12584461-connecting-custom-mcp-servers-in-chatgpt), accessed 2026-10-17 [^8]: [GitHub issue openai/codex #30418: Gmail connector should support multiple named accounts](https://github.com/openai/codex/issues/30418), accessed 2026-10-17 [^9]: [Google: Automatically forward Gmail messages to another account](https://support.google.com/mail/answer/10957?hl=en), accessed 2026-10-17 [^10]: [Microsoft: Turn automatic forwarding on or off in Outlook](https://support.microsoft.com/en-us/outlook/mail/turn-automatic-forwarding-on-or-off-in-outlook), accessed 2026-10-17 [^11]: [GitHub: google_workspace_mcp (README)](https://github.com/taylorwilsdon/google_workspace_mcp), accessed 2026-10-17 [^12]: [Command Mail: llms.txt with plans, scopes and connection options](https://commandmail.app/llms.txt), accessed 2026-10-17 [^13]: [Command Mail: Documentation](https://commandmail.app/en/docs), accessed 2026-10-17 [^14]: [Command Mail: For agents](https://commandmail.app/en/agents), accessed 2026-10-17 [^15]: [OpenAI: Managing app permissions in ChatGPT](https://help.openai.com/en/articles/20001495-managing-app-permissions-in-chatgpt), accessed 2026-10-17 --- Command Mail: one inbox for you and your AI agent. Start free: https://commandmail.app/login?signup=1 · Connect your agent: https://commandmail.app/en/agents · MCP: https://commandmail.app/mcp --- # Set up and use Command Mail. From your first address to a connected agent. You can start free and set everything up yourself. ## Quick start 1. **Account.** Create your account: [Start free](https://commandmail.app/login?signup=1). If you already have an account, [log in](https://commandmail.app/login) as usual. If you’d like help, request [personal setup](https://commandmail.app/en/anfragen?thema=einrichtung). 2. **Connect addresses.** Do it yourself or through your agent. Each address gets its own private inbound address. [Here’s how](https://commandmail.app/en/docs#weiterleiten). 3. **Connect your agent.** Add the MCP URL `https://commandmail.app/mcp` in Claude or ChatGPT, or use the Claude Code command. [Instructions by client](https://commandmail.app/en/docs#agent-verbinden). 4. **Approve permissions.** On the authorization page, you choose what your agent can do. [Scopes](https://commandmail.app/en/docs#scopes) and [Autonomy](https://commandmail.app/en/docs#autonomie) explain the levels. > Signing in with Google shares only your name, email address, profile picture link and account ID. Command Mail gets no access to Gmail, Drive or Calendar. ## Forward your addresses Command Mail is one inbox for multiple addresses and domains you own. Email arrives through forwarding, for example from Gmail or your own domain. Each address gets its own private inbound address. ### Connect an address Use the browser or CLI. Then set up forwarding with your existing provider to that address’s private inbound address. ``` commandmail mailboxes connect mara@beispiel.de ``` ### Your own domain Verify ownership of your domain with a DNS TXT record: first claim it, then verify it. ``` commandmail domains claim studio-beispiel.de # Add the TXT record at your DNS provider commandmail domains verify studio-beispiel.de ``` ### Import existing emails Import MBOX and EML exports or folders of EML files through the CLI. First connect the destination mailbox and log in to the CLI. For example: ``` commandmail import export.mbox --mailbox mara@beispiel.de --state archive ``` Import brings in existing messages. It does not set up continuous sync with your previous provider. ### Let your agent set it up If your agent has the scope `mailboxes`, it can handle these steps. For example, say: “Set up hallo@studio-beispiel.de in Command Mail.” It connects the address, claims the domain if needed, gives you the TXT record, verifies it and tells you where to forward email. > Receiving and sending are set up separately. From Pro, you set up sending for each sender address; you confirm ownership of your own domain via DNS. Changes in Command Mail do not automatically sync back to the original mailbox. ## Connect your agent All agents use the same operations through hosted MCP with OAuth, local MCP, CLI or REST. Every session starts with `capabilities`. This shows the agent the owner, scopes, mailboxes with their autonomy levels and its limits. ### Claude Web and desktop: add a custom connector with the URL `https://commandmail.app/mcp` and confirm access on the authorization page. ### ChatGPT Add `https://commandmail.app/mcp` as a remote MCP connector where your plan supports it. Confirm access on the authorization page. ### Claude Code ``` claude mcp add --transport http \ commandmail https://commandmail.app/mcp ``` ### Codex Local MCP through the CLI: `codex mcp add commandmail -- commandmail mcp`. Or use the hosted URL if your version of Codex supports remote MCP servers. ### Command EVE EVE and your bots connect through the hosted MCP URL and work with the permissions you grant them. ### Grok Through the xAI API: add `https://commandmail.app/mcp` as a remote MCP server. Otherwise, use the CLI or REST. ### Muse Through the MCP URL once Muse supports custom connectors. Until then, use REST. ### Hermes Use the hosted MCP URL if the runtime supports Streamable HTTP MCP. Otherwise, use the CLI or REST. This also applies to other agents. More on permissions, workflow and limits: [For agents](https://commandmail.app/en/agents). ## CLI The CLI is called `commandmail`. It works with scripts and scheduled tasks. Install with one line, then log in. Requires Node.js 18 or later. ``` curl -fsSL https://commandmail.app/install.sh | sh commandmail login ``` | Command | Purpose | | --- | --- | | `commandmail login` | log in | | `commandmail mailboxes connect ` | Connect an address | | `commandmail domains claim ` | Claim your own domain | | `commandmail domains verify ` | Verify a domain with a DNS TXT record | ## Custom views and the agent guide Use the **+** next to the inbox tabs to create a custom view, such as “Customer conversations”. The view uses a shared label and includes archived emails with that label. Creating a view does not move any email. Your agent receives our [guide to organizing and archiving](https://commandmail.app/skills/commandmail-organize/SKILL.md) with the CLI installation. Through MCP, it reads the guide with `get_organization_guide`, or through the CLI with `commandmail skill`. It also explains how to handle your exceptions, false positives, invoices and narrowly defined filters. ``` commandmail skill commandmail labels create --name "Kundenkommunikation" commandmail labels list --json commandmail inbox --label LABEL_ID --json ``` A new view starts with manual assignments. Enable automatic labels in settings. Your agent suggests spam and archive rules based on concrete examples; you enable them under “Automation & rules”. Marking something as spam alone does not activate a rule. In the morning brief, reported payment issues appear under “Review payments”, linked to the original email. A payment confirmation is not treated as an outstanding balance. Check the current balance with the provider; an email or risk score alone does not prove it. ## Scopes Scopes define what a key or connected session can do. Grant only what the agent needs. | Scope | Allowed | | --- | --- | | `mail:read` | Read, search and view without marking as read; overview and morning brief; sort within the autonomy level; view activity and undo its own actions; suggest rules; follow up | | `mail:draft` | Save drafts. Never sends or overwrites your edits | | `mail:send-approved` | Send the exact version you approved | | `mail:send` | Send and reply independently, each email with its own `sendKey`, within the daily limit (50 by default) | | `mailboxes` | Connect addresses, verify domain ownership with a TXT record, import old emails. Sending is set up for each sender address, from Pro | > Only a person in the browser can approve drafts, decide on suggestions, enable rules and change autonomy. Email content never grants permissions. Operator permissions such as `admin` are not available for customer keys. ## Autonomy Set the autonomy level for each mailbox: `read` → `suggest` → `rules` → `auto`. | Level | What the agent can do | | --- | --- | | `read` | Read only. It changes nothing, not even read markers | | `suggest` | Archiving, spam and deletion actions are saved as suggestions (response `{"suggested": true}`) and wait for you | | `rules` | As with `suggest`: agent actions remain suggestions. Only rules you have enabled take effect | | `auto` | Actions run and are logged, with at most 50 conversations per call. Supported sorting actions can be undone; sent emails cannot | ### Activity log Actions taken by agents and rules appear in the activity log and can be undone. Suggestions wait there for your decision. ## Let your agent reply For mailboxes such as customer support, a suitably configured agent can answer questions and send replies independently. It knows your offers, agreements and tone only from the information and sources you make available to it. Regular execution is configured separately in the agent. 1. **Grant permission.** On the authorization page, grant this agent `mail:send`. Agents that only sort email do not need it. 2. **Set a daily limit.** In settings under Automation; the default is 50 emails per day. 3. **Read the log.** Every sent email appears in the activity log. > Each email uses its own `sendKey`. Repeating the same request with the same key refers back to the same send attempt. Detected automated messages (RFC 3834, `Precedence: bulk`) are excluded from automatic replies. ## Morning brief The morning brief summarizes what arrived since yesterday and what is due by tomorrow: - Decisions waiting for you - Deadlines - Pending replies - Follow-ups that are due - What arrived since yesterday Expand it to the week with one button. Agents retrieve it with `morning_brief`. ## Risk assessment Each email receives percentage estimates for spam, fraud and phishing. The assessment reads text only. It never opens links or attachments. > Your account will be suspended in 24 hours > > Account service \ > > - Phishing: 94 % > - Fraud: 71 % > - Spam: 38 % ### Categories and markers Alongside risk assessment, Command Mail sorts email into General, Social, Promotions, Updates and Forums, and adds markers such as Appointment, Invoice, Deadline and Security. ## Frequently asked questions ### Does Command Mail read my Gmail account? No. Signing in with Google shares only your name, email address, profile picture link and account ID. Command Mail sees the emails you forward to your private inbound address. ### Can my agent reply and send independently? Yes, if you grant it `mail:send`. Every email is logged and the daily limit applies (50 by default). Without this permission, it saves drafts and sending follows your approval. [How to set it up](https://commandmail.app/en/docs#selbst-antworten) ### Will reloading send an email twice? Reloading does not create a new send attempt for the same approved draft version. If the outcome is unclear, check the recorded send status before starting a new send. ### Does an agent reply to newsletters? Never automatically: automated emails (RFC 3834, `Precedence: bulk`) receive no automatic replies. ### Is there an app? An installable web app with web push, plus Telegram notifications. Desktop apps for Mac and Windows are planned. ### Which languages does Command Mail support? German, English, Spanish, French and simplified Chinese. ### How much does it cost? You can start free with one address. Pro costs €19 per month as a founder price for the first 100 paying Pro accounts, as long as the subscription runs, and €29 per month for all later accounts; yearly billing gives you two months free. You can try Pro for 7 days for €1. Team costs €39 per person per month; Enterprise is priced on request. All prices include VAT. [See pricing](https://commandmail.app/en#preise) ### Who is behind Command Mail? Command Mail is a product of Fyn Labs LLC. Mathias Heinke answers inquiries and handles personal setup. --- Markdown mirror of https://commandmail.app/en/docs (as of 2026-10-17).