Skip to content

Connector builds / Build in-house vs hire a studio

Build it in-house, or hire us?

Build your connector in-house if you have an engineer with room on the roadmap, your API already uses OAuth, and you plan to grow the connector into a core product surface. Hire a studio if your roadmap is full, nobody on the team has taken a server through directory review, or you want it shipped without pulling people off the work that pays. There is a middle path: Connector Blueprint gives your engineers the reviewed design and a working prototype, and your team writes the rest.

Nine of our tools are listed in Anthropic’s Claude directory.

At a glance

In-house wins
Spare engineering room, OAuth already in place, a long-term surface
Studio wins
A full roadmap, no review experience, a date to hit
Middle path
Connector Blueprint: design and prototype, your build
Either way
The code is yours

Checked October 12, 2026

Side by side

The honest tradeoffs

FactorYour own teamModern Mustard Seed
Knowledge of your productDeep, and it stays in the buildingLearned from your docs and your team, written down in the handoff
Directory reviewLearned on your first submissionAlready done for our own connectors, with the counts shown on this page
Roadmap costEngineers come off other work while it shipsYour roadmap keeps moving
Long-term ownershipNatural: the people who built it maintain itThe code is in your repository; Connector Care is optional, and your team can take it over
When the protocol changesSomeone on your team tracks MCP, Claude and ChatGPTTracked for you under Connector Care, or handed to your team
Pricing shapeSalary and opportunity costA set package with a fixed scope, quoted in the conversation; changes included

When in-house is the right call

Keep it in-house when this is true.

I tell teams to build it themselves when most of these hold:

  • An engineer has real room on the roadmap, not room borrowed from a launch
  • Your API already supports OAuth, so the hardest plumbing exists
  • The connector will grow into a core product surface with its own roadmap, and you want the people who own it to have built it
  • Your team already ships against a public spec and enjoys reading one

When a studio is the right call

Hire it out when this is true.

Hire a studio when the roadmap is full and the connector keeps sliding, when nobody on the team has been through Anthropic’s or OpenAI’s review, or when competitors in your category already have a listing and you want yours on a date. What you buy is not typing speed. It is a design that already answers what reviewers ask: the tool names, the hints, the approval step, the privacy page they read first.

The middle path

Connector Blueprint: your team builds it.

If your engineers want to write it, Connector Blueprint gives them a head start that holds up in review: every tool named and mapped to your real endpoints and marked read-only or destructive, the auth pattern and the approval step for every write, the directory review checklist worked against your design, an eval set, and a working prototype on your public API. If you later decide we should build it, the Blueprint is credited in full toward a Connector Build.

Either way

Check these before you decide.

Whoever builds it, ask for evidence on each of these:

  • A server they have taken through a directory review, listed today
  • OAuth against real customer accounts, not a shared key
  • A write that waits for approval, with an audit trail
  • An eval set that runs before every release. We publish our own benchmark for agents doing small-business work, the Small Business Agent Bench, in the open
In review with Anthropic
4
In the official MCP Registry
15
Tools Claude can call
69

What we have shipped

Built and running, not a deck.

See every connector we have built for Claude

The package that fits

Connector Blueprint

If your team will build it, start with Connector Blueprint. If you want it built and submitted, Connector Build.

Your engineers start from a design that already answers what directory reviewers ask. Set scope, set price, quoted in the conversation, and changes to what we built are included.

Straight answers

What teams ask next.

Is it cheaper to build an MCP connector in-house?

Sometimes. If an engineer has real spare room and your API already uses OAuth, building in-house can cost less in cash. The cost to weigh is what that engineer would have shipped instead, and how many review rounds the first submission takes.

If we hire you, can our team take it over later?

Yes. The code lives in your repository and runs in accounts you control. Connector Care is optional, month to month, and your team can run the connector without us whenever you choose.

What does the Blueprint give our engineers that docs do not?

Decisions. Which endpoints become tools, how each is named, which ones are held for approval, how auth works for your API, and the review checklist worked against that design, plus an eval set and a prototype that runs.

Sources

I wrote this page for Modern Mustard Seed, and we sell the service it describes. Every fact about Anthropic, OpenAI or another company was read from the page linked here on October 12, 2026.

One next step

Send me your API docs.

Tell me your product and where your API reference lives. I will come back with the jobs your customers would do in Claude and ChatGPT, the tools I would build, and the package that fits. Managed for you, end to end, and the code is yours.

Claude is a product of Anthropic, and ChatGPT is a product of OpenAI. Modern Mustard Seed is an independent studio and is not affiliated with or endorsed by either.