Integration
MCP for HubSpot: the straight answer, and the pairing that works today.
HubSpot is where your customer actually lives: the contact, the company they belong to, the deal in flight, the lifecycle stage they've reached, the emails they've opened. So when you ask Claude for help with a win-back or a re-engagement cut, the advice comes back sensible and completely generic, because Claude has never seen one of your records. It can't tell you which customers went quiet after they became customers, or which MQLs stalled before sales ever called. If you searched "MCP for HubSpot" hoping to close that gap, this page owes you the honest version of what that search finds, including the part where the tool for the live reads isn't ours, and the part where the email work needs a different answer entirely.
What "MCP" actually means here.
MCP stands for Model Context Protocol, an open standard from Anthropic that lets Claude reach outside the chat window: a plugin system where the plugins give Claude real capabilities. Two facts before any pitch. First: HubSpot ships a first-party remote MCP server for its CRM. It's generally available, you connect it with OAuth so no token sits in a config file, and it covers contacts, companies, deals, and engagements for reading and writing, with campaigns, landing pages, and blog posts read-only. Use it. For reading your CRM live, it's the right tool and it isn't ours. Second: Orbit ships no HubSpot tools. Nothing in Orbit takes a HubSpot token. What Orbit is, is the lifecycle layer on top of the CRM: segment logic, lifecycle-stage review, a local email pipeline, all of it running without a credential from anyone. The full walk-through of what an MCP is and how it changes the work lives on the MCP for marketing page →.
What works beside HubSpot today.
No hand-waving. Each row names who provides it: the live CRM reads and writes come from HubSpot's own first-party MCP, the lifecycle layer comes from Orbit and runs locally with no credential, the export row says exactly how your data arrives, and the two rows that don't exist say so, along with which side the gap is on. We'd rather point you at the official tool than fake our own.
Read contacts, companies, deals, and engagements live
Works todayThrough HubSpot's own first-party remote MCP server, sign in with OAuth and Claude reads the CRM directly: the objects, their properties, the calls and emails and notes logged against them. This is HubSpot's tool, not Orbit's, and it's the right one: generally available, maintained by the people who run the API, no token pasted into a config file.
Write back to a CRM record
Works todayAlso HubSpot's MCP, its CRM coverage runs both ways, so a scoped property update is something you can genuinely ask for. What governs it is the OAuth scopes you approve at sign-in, not anything Orbit does. Our view on which writes belong in an unattended loop is further down, and it's a shorter list than the API allows.
Read campaigns, landing pages, and blog posts
Works todayHubSpot's MCP again, read-only for these three. Enough to see what marketing shipped and when; not enough to change it. For content someone else owns, read-only is the correct default rather than a limitation.
Design the segment, the ladder, and the measurement
Works todayOrbit's side, no key needed. The lifecycle skills carry the protocols: re-engagement cuts that don't cannibalise an active nurture, win-back ladders with the right gaps, holdout and sample-size maths so the result means something, deliverability strategy for the domain you send from. This is the half most HubSpot teams are actually short of.
Build and QA the email HubSpot will send
Works todayOrbit's local pipeline: MJML compiled to clean, dark-mode-safe HTML, a render gate that catches clipping and inbox breakage before anything ships, subject-line and preheader scoring. Runs on your machine with no credential from anyone.
Review lifecycle stages and workflow design
Partial: named limitOrbit's side, from an export. Pull contacts with their lifecycle stage, or the workflow definitions you want reviewed, and Claude runs the read a lifecycle lead would: the stage nobody ever leaves, the two workflows competing for the same people. Genuine analysis on your real records; the transport is a CSV.
Open HubSpot marketing emails and templates
Not available todayNot available from either side. HubSpot's MCP doesn't cover marketing email assets or templates, and Orbit has no HubSpot tools at all. The workaround is the plain one: build and QA the email in Orbit's local pipeline, then paste the HTML into HubSpot's editor yourself.
Call HubSpot's API from Orbit's own tools
Not available todayNot shipped, and not a credential Orbit takes. Every tool an MCP carries costs context in every conversation, and a HubSpot set hasn't earned its seat under that size budget yet. HubSpot's own server covers the live CRM in the meantime. If Orbit's tools ship later, this page changes then.
The short version: the live CRM reads exist today, from HubSpot, officially. The lifecycle reasoning and the email build exist today, from Orbit, locally. The gap is the marketing email surface, neither server touches it, and that's precisely the gap Orbit's local pipeline was built to fill.
The pairing in practice.
Connect HubSpot's MCP and install Orbit in the same Claude, and the two layers do the job a lifecycle owner currently walks by hand: one reading the CRM, one deciding what to do about it.
- Find the quiet customers, then build the win-back. Ask Claude to read the customers whose last engagement predates a threshold through HubSpot's MCP, then hand the shape to Orbit: how many emails the ladder should be, what the gaps between them are, which holdout proves it worked. The read is HubSpot's; the ladder is a lifecycle decision, and that's Orbit's.
- Audit the lifecycle stages nobody has looked at since 2023. Export contacts with their stage and dates, and Orbit's review names the stalls: the stage with a one-way door, the MQL definition that quietly became "filled in a form once", the transition that happens in bulk on the first of the month because a workflow does it rather than a human.
- Build the email HubSpot will actually send. HubSpot's MCP won't open a marketing email or a template, so this half never touches an API. Orbit compiles the MJML, runs the render gate, checks dark mode and the clipping threshold, scores the subject line, and you paste finished HTML into HubSpot's editor.
Nothing here needs a HubSpot credential handed to Orbit. The live reads run on HubSpot's own server under your OAuth sign-in; the reasoning and the build run on your machine.
How to set it up.
Two steps, and neither involves pasting a HubSpot token into anything.
- Connect HubSpot's own MCP for the live CRM. Add HubSpot's remote MCP server in Claude and sign in with OAuth. HubSpot's developer docs carry the current URL and the scope list. The sign-in is the credential, and the consent screen is where you decide how much of the CRM Claude can touch. Grant the reads first.
- Install Orbit for the lifecycle layer. It's free, and there is no HubSpot key step, because no Orbit tool would use one. The segmentation protocols, the lifecycle-stage review, the email pipeline, and the QA gate all run immediately.
If Orbit ships its own HubSpot tools later, the connection step will appear here, and it will start read-only, because that's the posture everywhere else Orbit connects.
What to never let an AI do unattended.
HubSpot's MCP writes as well as reads, which makes this worth saying out loud. A CRM is not a sandbox, the records are real people and the deals are real revenue, and two kinds of action have no business in an unattended loop no matter how good the tooling gets:
- Bulk edits. A mistaken filter or an off-by-one segment can rewrite a property across half your database before you notice, and the undo is a restore request. Ask for the change as a list you approve, not as a mutation that has already happened by the time you read the summary.
- Deletes. Archiving or deleting objects is the one action you can't comfortably reverse. It has no place in an autonomous loop. Full stop.
Sends are the third one people worry about, and here the answer is structural rather than a matter of discipline: HubSpot's MCP doesn't cover marketing email, so firing a campaign isn't on the table through this path at all. Sends stay in HubSpot's UI, under your hand, where a human presses the button.
None of this is Orbit gating anything. Orbit has no HubSpot tools to gate with. It's the posture we'd take if it did, and the one you can enforce today at HubSpot's OAuth consent screen by granting the read scopes and leaving the writes alone until you have a specific reason.
What's deepest on Braze, and why we say so.
Orbit's deepest single-platform integration is Braze, where the write path runs end to end. A handful of its tools are built around concepts that only exist in Braze, Canvas flows, Content Blocks, its media library, and those stay Braze-named on purpose. HubSpot has its own automation model, its own email tooling, its own object graph; pretending they map one-to-one onto Braze's would be a worse version of both.
For a HubSpot team the trade barely bites, because the portable work is the work: reading the funnel, designing the segment, running the numbers, building the template. And if your sends ever graduate to a lifecycle-native platform, the tooling and the discipline travel with you.
The half that isn't about the API at all.
Reading the CRM is one half. The reason to bother is what Claude does with the reading. Orbit ships 88 lifecycle skills, structured protocols for the actual jobs: composing an email, laddering a win-back, scoring a subject line, running an audit, reasoning about a funnel. Those are platform-agnostic; they work the same whether the CRM underneath is HubSpot or anything else.
Underneath sit 136 tools, an MJML pipeline that compiles to clean, dark-mode-safe HTML, a render-and-QA gate that catches the clipping and inbox breakage before you send, and the calculators a real test needs: sample size, holdout, replenishment timing. HubSpot holds the customer. Orbit is the practitioner sitting next to it, helping you decide what to do and building the email that does it.
Is this for you?
This is for the lifecycle or RevOps marketer who runs on HubSpot and is doing two jobs at once: reading the funnel to work out who to talk to, then building the email that talks to them. If that's your week, the working setup is HubSpot's own MCP for the CRM and Orbit for the judgement and the build on top. It's a weaker fit if what you wanted was a single Orbit tool that reads your HubSpot portal; that doesn't exist, and we won't pretend otherwise. What you get instead is two real things that work this afternoon, and a straight answer about which is which, rarer than it should be.
Try it.
Every Orbit plan includes every skill and tool, on every ESP. Install it, connect HubSpot's own MCP beside it, and the whole pairing runs the same afternoon, no token pasted anywhere, nothing to revoke later.
Frequently asked.
What is an MCP for HubSpot?
An MCP, Model Context Protocol, an open standard from Anthropic, gives Claude real tools instead of general knowledge. For HubSpot there are two honest answers. HubSpot itself ships a first-party remote MCP server for its CRM: you sign in with OAuth and Claude reads and writes contacts, companies, deals, and the engagements attached to them, and reads campaigns, landing pages, and blog posts. Orbit ships no HubSpot tools today; nothing in Orbit takes a HubSpot token. What Orbit is, is the lifecycle layer on top: segment logic, lifecycle-stage and workflow review, and a local email pipeline, none of it needing a credential from anyone. The pairing that works today is HubSpot's MCP for the live CRM and Orbit for the lifecycle craft. Orbit is not an official HubSpot product; it is an independent MCP, and only the first half of that pairing is HubSpot's own.
Is there an official HubSpot MCP?
Yes, and it is the right tool for live CRM data. HubSpot hosts a remote MCP server, generally available, connected with OAuth, so no token ever sits in a config file and the sign-in is the secret. It covers CRM objects and engagements for both reading and writing, plus read-only access to campaigns, landing pages, and blog posts. What it does not cover is marketing email assets and templates. So the server is excellent at the records and silent on the sends, which is worth knowing before you plan your week around it.
Does Orbit connect to HubSpot directly?
No. Orbit ships no HubSpot tools and takes no HubSpot credential: no private app, no access token, no OAuth. An earlier version of this page said otherwise, in some detail, and it was wrong; correcting it matters more than looking consistent. The reason for the gap is a size budget: every tool an MCP carries costs context in every conversation, and a HubSpot set has not yet earned its seat against the tools people use daily. HubSpot's own MCP already covers the live CRM, officially. If Orbit's tools ship later, this page changes then.
What does Orbit add on top of HubSpot's MCP?
The judgement, and the email. HubSpot's MCP reads the CRM; Orbit knows what the reading means for your lifecycle program: how to cut a re-engagement segment that will not cannibalise the nurture, where a win-back ladder should start, what to hold out so you can prove it worked. Then it does the part HubSpot's server leaves out entirely: a local email pipeline that compiles MJML to dark-mode-safe HTML, a render gate that catches clipping and inbox breakage before you send, subject-line and preheader scoring, and deliverability strategy for the domain you send from. All of it runs on your machine without a HubSpot credential, because none of it is a HubSpot read.
Can Orbit reason about HubSpot lifecycle stages and workflows?
Yes, from an export. Pull your contacts with their lifecycle stage, or the workflow definitions you want a second opinion on, and hand the file to Claude. Orbit brings the practitioner's read: the stage everyone enters and nobody leaves, the two workflows enrolling the same people into competing messages, the stage definitions that stopped meaning what they meant when someone wrote them three years ago. The analysis on your real records is genuine. The transport is a CSV instead of a credential, and we would rather name that than imply a connection that does not exist.