Connect your AI to Help Scout

What your AI can do with Help Scout

Once connected, these tools appear automatically in your AI assistant:

• list_mailboxesread-only

  List the mailboxes of the account.

• list_conversationsread-only

  List conversations, optionally filtered by mailbox and status.

• get_conversationread-only

  Get a conversation with its threads (messages).

• list_conversation_threadsread-only

  List a conversation's threads on their own, paginated — what the customer wrote, what each agent replied, and the internal notes. Use this rather than get_conversation when a long exchange needs reading past its first page.

• search_conversationsread-only

  Search conversations by their content, subject, customer or tag, using Help Scout's own query syntax — for example `subject:"remboursement"` or `tag:urgent`. Narrower than listing a mailbox and reading through it.

• add_notewrite · needs approval

  Add an internal note to a conversation. Only the team sees it — nothing reaches the customer. This is the safe way for an assistant to leave a summary or a finding on a ticket.

• reply_to_conversationwrite · needs approval

  Write a reply to the customer. It is created as a DRAFT by default, sitting in Help Scout for a human to read and send — because a published reply is emailed to the customer immediately and cannot be recalled. Pass `draft: false` only when the customer has asked for the reply to go out now. For something only the team should see, use add_note.

• schedule_replywrite · needs approval

  Schedule an existing draft reply to go out at a chosen time. Applies to a draft thread created by reply_to_conversation — take its id from the result. Scheduling again overrides the previous time.

• create_conversationwrite · needs approval

  Open a conversation in a mailbox with a first thread — logging a phone call, recording something a customer said elsewhere. Creating it does not email the customer unless `autoReply` is set. For history brought over from another tool, set `imported` so Help Scout generates no outgoing mail or notification at all.

• update_conversationwrite · needs approval

  Change a conversation's subject, status, owner or mailbox. Assign to a person or a team with `assignTo`, or hand it back to the queue with `unassign`.

• set_conversation_tagswrite · needs approval

  Set the tags of a conversation. THIS REPLACES THE WHOLE SET: every tag missing from the list is removed, and an empty list clears them all. There is no way to add one tag on its own — read the conversation's current tags first with get_conversation, then send them back together with the new one.

• list_customersread-only

  List or search customers, by email, name or the organization they belong to.

• get_customerread-only

  Get one customer's profile, with their emails, phones and organization.

• create_customerwrite · needs approval

  Create a customer profile.

• update_customerwrite · needs approval

  Update a customer's profile fields. Only what you pass is changed; everything else is left alone. Emails and phones are managed on their own and are not touched here.

• get_mailboxread-only

  Get one mailbox, with its sending address and settings.

• list_saved_repliesread-only

  List a mailbox's saved replies — the wording the team has already agreed on. Worth reading before drafting an answer, so a reply matches how the team actually writes.

• list_usersread-only

  List the agents of the account.

• get_userread-only

  Get one agent by id.

• get_current_userread-only

  Get the agent the connection belongs to — who Help Scout thinks is acting. Worth knowing before assigning or signing anything.

• list_teamsread-only

  List the teams of the account. A team can own a conversation, like an agent can.

• list_team_membersread-only

  List the agents belonging to a team.

• list_tagsread-only

  List the tags of the account. Read this before set_conversation_tags, so an existing tag is reused rather than a near-duplicate created.

• get_tagread-only

  Get one tag by id.

• list_organizationsread-only

  List the organizations — the companies customers are grouped under, with their domains, website and notes.

• get_organizationread-only

  Get one organization by id.

• get_ratingread-only

  Get one satisfaction rating: the score (great, okay, not_good), the customer's comment, the agent whose reply was rated, and the conversation it belongs to. The id is the rating's own, not the conversation's.

• get_company_reportread-only

  The company-wide overview for a period: customers helped, conversations closed, replies sent, replies per day and per user, each with the same figures for a previous period and the delta between them. Start here, then drill into whichever number looks wrong.

• get_company_customers_helpedread-only

  How many distinct customers the team helped, plotted over the period. `viewBy` sets the resolution: day up to 60 days, week up to a year, month beyond 61 days.

• get_company_drilldownread-only

  The actual conversations behind a company figure, one page at a time. `range` says which figure — replies, firstReplyResolved, resolved and the rest — and `rangeId` narrows it further where the range takes one.

• get_conversations_reportread-only

  Conversation volume for a period: how many came in, how many were closed, how many are still open, against the previous period.

• get_conversations_newread-only

  New conversations over time — the inbound curve, at the resolution `viewBy` asks for.

• get_conversations_busy_timesread-only

  When conversations actually arrive, as a grid of weekday against hour. This is the report that answers staffing questions rather than performance ones.

• get_conversations_received_messagesread-only

  Messages received over time, which is not the same as conversations: one thread can carry many.

• get_conversations_volume_by_channelread-only

  Volume split by channel — email, chat, phone — over time. Where the load comes from.

• get_conversations_drilldownread-only

  The conversations counted by the conversations report, listed one page at a time.

• get_conversations_new_drilldownread-only

  The conversations counted as new, listed one page at a time.

• get_conversations_field_drilldownread-only

  The conversations behind one tag, saved reply, workflow or customer. `field` says which of the four, `fieldId` says which one of them.

• get_productivity_reportread-only

  The team's productivity for a period: first response time, response time, resolution time, replies sent, resolved count, against the previous period. `officeHours` measures elapsed time against the account's business hours instead of the clock.

• get_productivity_first_response_timeread-only

  First response time over the period — how long a customer waits for the first human reply.

• get_productivity_response_timeread-only

  Response time over the period, counting every reply and not only the first.

• get_productivity_resolution_timeread-only

  Resolution time over the period — how long a conversation stays open before it is closed.

• get_productivity_replies_sentread-only

  Replies sent over the period, at the resolution `viewBy` asks for.

• get_productivity_resolvedread-only

  Conversations resolved over the period, at the resolution `viewBy` asks for.

• get_happiness_reportread-only

  The satisfaction score for a period: how many ratings were great, okay and not good, with the same breakdown for the previous period.

• get_happiness_ratingsread-only

  The individual satisfaction ratings behind the score, with the customer's comment. `rating` filters to great, ok or not-good. This is where you read what an unhappy customer actually said.

• get_user_reportread-only

  One agent's figures for a period: conversations handled, replies sent, response and resolution times, happiness, against the previous period.

• get_user_customers_helpedread-only

  How many distinct customers one agent helped, plotted over the period.

• get_user_repliesread-only

  One agent's replies sent over the period, at the resolution `viewBy` asks for.

• get_user_resolutionsread-only

  One agent's resolved conversations over the period, at the resolution `viewBy` asks for.

• get_user_happinessread-only

  One agent's satisfaction score for the period, against the previous one.

• get_user_ratingsread-only

  The individual ratings one agent received, with the customer's comment. Filter with `rating` to read only the bad ones.

• get_user_conversation_historyread-only

  The conversations one agent handled over the period, listed with their reply count, response time and resolution time. Sort with `sortField` to bring the slowest to the top.

• get_user_drilldownread-only

  The conversations behind one agent's figures, listed one page at a time.

• get_user_chat_reportread-only

  One agent's chat figures for the period: chats handled, response and resolution times.

• get_chat_reportread-only

  The team's chat figures for the period, against the previous one.

• get_email_reportread-only

  The team's email figures for the period, against the previous one.

• get_phone_reportread-only

  The team's phone figures for the period, against the previous one.

• get_docs_reportread-only

  How the Docs knowledge base performed over the period: views, searches, and the searches that returned nothing — the list of articles your customers wanted and did not find.

Example prompts

Just ask your AI in plain language — it picks the right Help Scout tool:

• “List the mailboxes of the account — from Help Scout, in plain language.”
• “List conversations, optionally filtered by mailbox and status — from Help Scout, in plain language.”
• “Get a conversation with its threads (messages) — from Help Scout, in plain language.”
• “List a conversation's threads on their own, paginated — what the customer wrote, what each agent replied, and the internal notes. Use this rather than get_conversation when a long exchange needs reading past its first page — from Help Scout, in plain language.”

How it works

1. Create your Proxoria account and add your MCP endpoint to Claude or ChatGPT.
2. Connect Help Scout with your API key — each user connects their own account, credentials encrypted at rest.
3. Your AI instantly gets the Help Scout tools. Write actions wait for your approval in the dashboard before anything changes.

Frequently asked questions

Does Help Scout have an official MCP server?

You don't need one. Proxoria exposes Help Scout to your AI through the Model Context Protocol, so you can use it from Claude, ChatGPT, Mistral or any MCP-compatible client — no official Help Scout MCP server required.

What can my AI do with Help Scout through Proxoria?

It gets 59 tools: list_mailboxes, list_conversations, get_conversation, list_conversation_threads, search_conversations, add_note, reply_to_conversation, schedule_reply, create_conversation, update_conversation, set_conversation_tags, list_customers, get_customer, create_customer, update_customer, get_mailbox, list_saved_replies, list_users, get_user, get_current_user, list_teams, list_team_members, list_tags, get_tag, list_organizations, get_organization, get_rating, get_company_report, get_company_customers_helped, get_company_drilldown, get_conversations_report, get_conversations_new, get_conversations_busy_times, get_conversations_received_messages, get_conversations_volume_by_channel, get_conversations_drilldown, get_conversations_new_drilldown, get_conversations_field_drilldown, get_productivity_report, get_productivity_first_response_time, get_productivity_response_time, get_productivity_resolution_time, get_productivity_replies_sent, get_productivity_resolved, get_happiness_report, get_happiness_ratings, get_user_report, get_user_customers_helped, get_user_replies, get_user_resolutions, get_user_happiness, get_user_ratings, get_user_conversation_history, get_user_drilldown, get_user_chat_report, get_chat_report, get_email_report, get_phone_report, get_docs_report. For example, list the mailboxes of the account.

Is the Help Scout connector read-only?

No — it can also make changes. Write actions go through Proxoria's confirmation guardrail by default: your AI proposes them, and you approve each one from your dashboard before anything is written to Help Scout.

How do I connect Help Scout to my AI?

Create a Proxoria account, connect Help Scout with your API key, and add your Proxoria MCP endpoint to Claude or ChatGPT. Your AI instantly gains the Help Scout tools.