Gmail MCP Service

This page documents every Gmail action available through AVELIN-MCP and what the model can execute with each action.

Gmail operations are part of AVELIN's Enterprise Platform pillar and execute through Cross-Model MoE cascading. Email workflows benefit from intelligent tier selection — routine operations use fast tier while complex drafting uses the pro tier or the coding tier tiers, delivering up to 87% cost savings on agentic tasks.

Service Purpose

Gmail MCP gives the model practical email operations: reading, searching, drafting, replying, organizing, filtering, and attachment handling.

Full Action List (Implemented)

  1. list_emails
  2. list_inbox
  3. get_email
  4. search_emails
  5. send_email
  6. create_draft
  7. list_labels
  8. reply_to_email
  9. trash_message
  10. archive_message
  11. unarchive_message
  12. star_message
  13. unstar_message
  14. get_unread_count
  15. mark_as_spam
  16. remove_from_spam
  17. list_blocked_senders
  18. block_sender
  19. block_domain
  20. unblock_sender
  21. list_auto_archived_senders
  22. auto_archive_sender
  23. auto_archive_domain
  24. unauto_archive_sender
  25. mark_as_read
  26. mark_as_unread
  27. add_label
  28. remove_label
  29. list_attachments
  30. get_attachment
  31. get_profile
  32. list_aliases

Action-by-Action Business Description

list_emails

  • What model can do: list recent messages across all mail.
  • Typical inputs: max_results, optional query, optional page_token.
  • Business outcome: broad email visibility for audits and context gathering.

list_inbox

  • What model can do: list inbox-only messages.
  • Typical inputs: max_results, optional query, optional page_token.
  • Business outcome: supports active inbox triage workflows.

get_email

  • What model can do: fetch one email or batch of emails.
  • Typical inputs: message_id or message_ids.
  • Business outcome: full message context for response drafting and decision support.

search_emails

  • What model can do: search with Gmail query syntax.
  • Typical inputs: query, max_results, optional page_token.
  • Business outcome: finds relevant communications quickly by topic, sender, or timeframe.

send_email

  • What model can do: send a new outbound email.
  • Typical inputs: to, subject, body.
  • Business outcome: converts decisions into immediate communication actions.

create_draft

  • What model can do: create draft messages without sending.
  • Typical inputs: to, subject, body.
  • Business outcome: enables review workflows before final send.

list_labels

  • What model can do: retrieve available label taxonomy.
  • Typical inputs: none.
  • Business outcome: supports consistent mailbox organization.

reply_to_email

  • What model can do: reply to an existing thread.
  • Typical inputs: message_id, body, optional reply_all.
  • Business outcome: preserves thread continuity and faster response handling.

trash_message

  • What model can do: move one or many messages to trash.
  • Typical inputs: message_id or message_ids.
  • Business outcome: inbox cleanup and noise reduction.

archive_message

  • What model can do: archive one or many messages.
  • Typical inputs: message_id or message_ids.
  • Business outcome: keeps inbox focused while preserving records.

unarchive_message

  • What model can do: return archived messages to active inbox.
  • Typical inputs: message_id or message_ids.
  • Business outcome: restores items for active handling when needed.

star_message

  • What model can do: star one or many messages.
  • Typical inputs: message_id or message_ids.
  • Business outcome: flags priority communication for follow-up.

unstar_message

  • What model can do: remove star flags.
  • Typical inputs: message_id or message_ids.
  • Business outcome: keeps priority queue current and meaningful.

get_unread_count

  • What model can do: get unread message count.
  • Typical inputs: none.
  • Business outcome: quick workload signal for communication backlog.

mark_as_spam

  • What model can do: mark messages as spam.
  • Typical inputs: message_id or message_ids.
  • Business outcome: reduces unwanted communication noise and risk.

remove_from_spam

  • What model can do: restore mistakenly flagged messages.
  • Typical inputs: message_id or message_ids.
  • Business outcome: prevents loss of legitimate business communication.

list_blocked_senders

  • What model can do: list sender/domain block filters.
  • Typical inputs: none.
  • Business outcome: visibility into current inbound filtering rules.

block_sender

  • What model can do: create sender-level block filter.
  • Typical inputs: email.
  • Business outcome: suppresses recurring unwanted messages from one sender.

block_domain

  • What model can do: create domain-level block filter.
  • Typical inputs: domain.
  • Business outcome: broad protection from unwanted domain traffic.

unblock_sender

  • What model can do: remove an existing block filter.
  • Typical inputs: filter_id.
  • Business outcome: restores communication channels when policy changes.

list_auto_archived_senders

  • What model can do: list auto-archive rule set.
  • Typical inputs: none.
  • Business outcome: governance visibility on inbox automation.

auto_archive_sender

  • What model can do: auto-archive and mark read for a sender.
  • Typical inputs: email.
  • Business outcome: reduces inbox clutter from low-priority senders.

auto_archive_domain

  • What model can do: auto-archive and mark read for a domain.
  • Typical inputs: domain.
  • Business outcome: streamlines high-volume low-priority traffic.

unauto_archive_sender

  • What model can do: remove auto-archive filter.
  • Typical inputs: filter_id.
  • Business outcome: recovers active inbox visibility for previously filtered traffic.

mark_as_read

  • What model can do: mark messages as read.
  • Typical inputs: message_id or message_ids.
  • Business outcome: keeps unread queue accurate after handling.

mark_as_unread

  • What model can do: mark messages as unread.
  • Typical inputs: message_id or message_ids.
  • Business outcome: re-queues items needing later action.

add_label

  • What model can do: apply a label to a message.
  • Typical inputs: message_id, label_id.
  • Business outcome: better categorization and retrieval.

remove_label

  • What model can do: remove a label from a message.
  • Typical inputs: message_id, label_id.
  • Business outcome: maintains clean information taxonomy.

list_attachments

  • What model can do: list attachment metadata in a message.
  • Typical inputs: message_id.
  • Business outcome: helps users identify required files quickly.

get_attachment

  • What model can do: retrieve attachment content.
  • Typical inputs: message_id, attachment_id.
  • Business outcome: enables downstream document workflows and reviews.

get_profile

  • What model can do: retrieve account profile information.
  • Typical inputs: none.
  • Business outcome: confirms account identity and mailbox scope.

list_aliases

  • What model can do: list configured send-as aliases.
  • Typical inputs: none.
  • Business outcome: supports correct sender identity in outbound communication.

Model-Achievable Gmail Scenarios

1) Morning inbox triage

  • Trigger: start of day.
  • Actions used: get_unread_count, list_inbox, star_message, archive_message.
  • Outcome: priority queue established.
  • Business benefit: quicker focus on high-value communication.

2) Fast response handling

  • Trigger: urgent inbound request.
  • Actions used: get_email, reply_to_email, mark_as_read.
  • Outcome: thread response sent quickly.
  • Business benefit: better responsiveness and stakeholder trust.

3) Draft-before-send policy

  • Trigger: sensitive outbound message.
  • Actions used: create_draft.
  • Outcome: reviewable draft.
  • Business benefit: reduced communication risk.

4) Targeted message retrieval

  • Trigger: user needs historical communication proof.
  • Actions used: search_emails, get_email.
  • Outcome: exact thread retrieved.
  • Business benefit: faster evidence and context retrieval.

5) Bulk inbox cleanup

  • Trigger: inbox saturation.
  • Actions used: list_inbox, archive_message, trash_message.
  • Outcome: inbox normalized.
  • Business benefit: lower cognitive load.

6) Anti-noise sender controls

  • Trigger: repeated low-value sender traffic.
  • Actions used: block_sender, block_domain, list_blocked_senders.
  • Outcome: recurring noise suppressed.
  • Business benefit: improved signal-to-noise ratio.

7) Low-priority stream automation

  • Trigger: newsletters or automated updates flooding inbox.
  • Actions used: auto_archive_sender, auto_archive_domain, list_auto_archived_senders.
  • Outcome: auto-managed low-priority stream.
  • Business benefit: better focus on critical communications.

8) Spam recovery

  • Trigger: important mail flagged as spam.
  • Actions used: search_emails, remove_from_spam.
  • Outcome: message restored.
  • Business benefit: prevents missed business-critical messages.

9) Label-based processing queue

  • Trigger: team uses labels for process stages.
  • Actions used: list_labels, add_label, remove_label.
  • Outcome: messages mapped to workflow stage.
  • Business benefit: stronger process visibility.

10) Attachment-first workflow

  • Trigger: incoming message with required document.
  • Actions used: list_attachments, get_attachment.
  • Outcome: required file extracted.
  • Business benefit: faster handoff to downstream work.

11) Alias-aware outbound communication

  • Trigger: user must send from specific identity.
  • Actions used: list_aliases, create_draft, send_email.
  • Outcome: message prepared and sent with correct profile context.
  • Business benefit: better brand and role consistency.

12) Re-open archived action item

  • Trigger: archived thread becomes active again.
  • Actions used: search_emails, unarchive_message, mark_as_unread.
  • Outcome: thread returns to active queue.
  • Business benefit: avoids lost follow-ups.