Calendly MCP with Claude: brief every meeting with competitor and AI visibility data

The Calendly connector in Claude's directory lists 36 tools: 24 read, 9 write, and one each classed as transactional, admin and unclassified. Claude can list and read scheduled events and their invitees, list event types and their available times, read your availability, create or update event types, create a single-use scheduling link, cancel an event, and read routing form submissions.

Calendly's blog says you can use Calendly MCP on any Calendly plan, including the free tier. Its developer docs say the server is fully hosted by Calendly, issues the scopes mcp:scheduling:read and mcp:scheduling:write, and works only with clients that support Dynamic Client Registration. If your client asks you to paste a client id and secret, it will not connect today.

Calendly is the calendar side. It can tell Claude who is booked on Thursday at 2 p.m. and what the meeting is called. It has no view of what a competitor shipped this week or what the answer engines say about your category.

Linkeddit is the market side. Compete tracks competitor moves with a priority on each, and Answer Radar records which brands the five engines name. Both come with one plan at $60 a month. Linkeddit writes no advice and books nothing. Claude reads both connectors, and you decide what to do.

Ask Claude to list this week's events with meetings-list_events, read the invitees on the ones you name, and then pull the latest competitor brief from Linkeddit. It can match a company in the invite to a tracked competitor, or tell you there is none. Calendly's own tools return the event and invitee details; they do not say which competitor a prospect uses, so the prompt must name the competitors that matter.

On Linkeddit's test project, plausible.io, the latest weekly brief covers 5 monitored competitors and 19 signals: 2 high, 9 worth watching and 8 low. The high ones are Fathom launching an AI assistant called Luna and Matomo publishing critical PHP security guidance. A call with someone comparing those two is better with that in hand than without it.

Scheduling data raises its own questions. A viewer on a YouTube video about Calendly asked "does anyone know how easy it is to get MI on % of shows, cancellations and how many appointments happened within a timeframe". That is a question for the Calendly side, which Linkeddit does not answer. Linkeddit adds the outside half: what the competitor did that week.

Yes, as a prompt you run, not as something Claude does by itself. Claude acts when it is started, so you run a saved prompt each Monday or start it from a scheduler you already use. It lists the week's events from Calendly and attaches the week's top competitor moves and answer gaps from Linkeddit.

Linkeddit's side does run on its own clock. On the test project all five monitors are set to weekly, each with a next run of 2026-11-04, so the data is ready when the prompt runs. For gaps, answer_actions_get returned two open cards for the question 'Umami vs Matomo for self-hosted analytics': one where a competitor holds the answer, with 12 competitor citations, and one content gap.

Yes, but it costs credits, so it is a step you approve. Before a review call, Claude can ask Linkeddit to re-check the questions a published change targeted. Linkeddit returns a credit estimate first, and Claude waits for your yes before it starts anything.

Then Claude reads the result and puts one line in the brief: whether the engines now name you. If the run is partial, the brief says so. Calendly's part is only to find the review meeting and its invitees, and optionally to make a single-use scheduling link with scheduling_links-create_single_use_scheduling_link if you want to book a follow-up.

Claude reads Calendly through your own sign-in, so it can see what your account can see. Limit the prompt to what it needs, for example this week's events, and keep it read-only unless you ask for a change. Calendly's tools include cancelling events and creating invitees, so tell Claude not to use them in a briefing prompt.

Then read the brief. Check that each competitor claim has a source link and that nothing was guessed about a guest's company. On the test project the Plausible monitor itself had analysed nothing yet, and a good brief says so.