Tool design
Six to twelve tools, each mapped to an endpoint your product already has and each marked read-only or destructive. The tool names and descriptions are what the assistant reads, so we write them for the assistant.
MCP server development / Claude and ChatGPT
An MCP server development company, from tool design to the directory.
Modern Mustard Seed is an MCP server development company in Kalispell, Montana. We design the tools, build the server on your real API, add OAuth sign-in and an approval step before every write, test it against an eval set, publish it to the official MCP Registry and submit it to Anthropic’s Claude directory and to ChatGPT. We build and run our own servers the same way. The code lives in your repository.
Nine of our tools are listed in Anthropic’s Claude directory.
- Listed in Anthropic’s Claude directory
- 9
- In review with Anthropic
- 4
- In the official MCP Registry
- 18
- Tools Claude can call
- 69
What the work includes / Design to care
Eight things an MCP server needs before it ships.
A server that answers a tool call is the easy part. A server a directory will list and a customer will trust is the list below. We do all of it, and put the scope in writing before we start.
Sign-in for each customer
OAuth sign-in, so each customer connects their own account. Where a server needs no account, it says so plainly.
An approval step on every write
A tool that changes something drafts the change first and returns a confirm token. Nothing is written until the customer says yes in the chat, and the action lands in an audit log.
An eval set
An eval set your team runs before every release, so a change to the server or the model shows up before a customer sees it.
Docs and a privacy page
The two pages a directory reviewer reads first, written against the published review rules and linked from the listing.
Registry and directory submissions
The server published to the official MCP Registry and submitted to Anthropic’s Claude directory and to ChatGPT. The listing decision belongs to Anthropic and OpenAI, and we say so.
Hosting and care
Monitoring with an alert the day a tool fails, updates when the protocol or the models change, and the listings kept current. Optional, month to month, and changes to what we built are included.
The code is yours
It lives in your repository and runs in accounts you control. We can run it for you end to end, and your team can run it without us whenever you choose.
How it runs / From your docs to the directory
Built on your API, not a guess.
We read your docs
Your public API reference and your product pages: which jobs your customers come to you for, and which endpoints already do them.
We design the tools
Six to twelve tools mapped to your real endpoints. Each one is marked read-only or destructive, and every write gets an approval step.
We build and test it
OAuth sign-in, the tools, an eval set, and every published directory rule checked before anything is submitted.
We submit it and keep it working
Docs and privacy pages written, the registry publication and both directory submissions filed, and the code in your repository.
The proof / Live servers, checked October 11, 2026
Fourteen MCP servers we built and run.
The table lists the fourteen connectors on our Built for Claude page. Each one answered an MCP initialize on its live URL on October 11, 2026, and each tool list is what the server returned. The first listings went live in Anthropic’s directory on October 12, 2026. Eighteen servers are published under our names in the official MCP Registry.
| Server and directory status | What it does | Tools and sign-in |
|---|---|---|
| Site Check (opens in a new tab)Listed in Anthropic’s Claude directory | Speed, security, domain, email, search basics, AI search readiness and contact paths, with a fix for every problem, plus a local presence score. | 4 tools. No sign-in. |
| Dry Run (opens in a new tab)Listed in Anthropic’s Claude directory | Simulates a year of cash for a small services business ten thousand times: the odds of running short, the week it most likely happens, what causes it, and which single change takes the most risk off. | 3 tools. No sign-in. |
| Idea to Product Sequencer (opens in a new tab)Listed in Anthropic’s Claude directory | Scores an idea on ten assumptions, picks the cheapest test with pass and kill lines written in advance, cuts the MVP, dates the build, and makes the push, park or kill call against the numbers. | 7 tools. No sign-in. |
| Own It (opens in a new tab)Listed in Anthropic’s Claude directory | Maps who controls the domain, DNS, website platform, email, tracking tags, booking tools and profiles, how hard each would be to leave, and the first step to take control. | 3 tools. No sign-in. |
| Package Pricing Lab (opens in a new tab)In review for Anthropic’s Claude directory | Turns a service business’s costs, capacity and deliverables into three fixed packages with the margin math shown, stress tests them against scope creep and a slow month, and writes the scope lines. | 4 tools. No sign-in. |
| Domain Health (opens in a new tab)Listed in Anthropic’s Claude directory | SPF, DKIM, DMARC, BIMI, MTA-STS and TLS-RPT, the certificate and TLS versions, security headers, and the registry record with its locks and DNSSEC, each with a fix and the rule behind it. | 5 tools. No sign-in. |
| AI Search Kit (opens in a new tab)Listed in Anthropic’s Claude directory | AI crawler rules, structured data, llms.txt and name, address and phone consistency, with LocalBusiness JSON-LD and llms.txt written for you. Every rule cites its public source. | 5 tools. No sign-in. |
| Local Trades Kit (opens in a new tab)In review for Anthropic’s Claude directory | Maps the ZIPs, towns and homes inside a service area, shows busy and slow months by trade and state, times review requests within the rules, and lists the pages a trade website needs. | 5 tools. No sign-in. |
| Call Desk (opens in a new tab)Listed in Anthropic’s Claude directory | Validates a US or Canadian number, finds its time zone, says whether a call is allowed right now under federal and state calling hours, and gives recording consent, AI voice disclosure and toll-fraud risk, every rule cited. | 5 tools. No sign-in. |
| Voice Agent Linter (opens in a new tab)Listed in Anthropic’s Claude directory | Checks a voice agent’s prompt and its Vapi, Retell, ElevenLabs or OpenAI Realtime config against rules learned on live calls, with the severity, exact line and fix for every finding. | 3 tools. No sign-in. |
| Small Business Funding Finder (opens in a new tab)In review for Anthropic’s Claude directory | Searches live federal grant listings on Grants.gov by keyword, applicant type, category, agency and deadline, opens any listing in full, and decodes who is eligible. Every result links to its official listing. | 4 tools. No sign-in. |
| Chief Dreamer (opens in a new tab)Live now, add it by URL | Gives Claude a private place to keep your ideas: what you dream, how well it fits your life right now, the smallest real test, and what is due next, remembered across every chat. You sign in with your email, and you can export or delete everything. | 12 tools. Signs in with the person’s own account. |
| Cross + Covenant Bible (opens in a new tab)In review for Anthropic’s Claude directory | Read and search the World English Bible and the King James Version, follow a reading plan day by day, and open the daily devotional. Every verse comes from public-domain text. | 5 tools. No sign-in. |
| Church Year (opens in a new tab)Listed in Anthropic’s Claude directory | Liturgical seasons and colors, feast days, Easter and Orthodox Pascha, and the three-year Sunday readings with public-domain Bible text, for pastors, worship planners and anyone who keeps the Christian year. | 4 tools. No sign-in. |
Owner’s Desk
Our own OAuth 2.1 connector. It reaches the owner’s Stripe and Google accounts, every write waits for a confirm token, and every action lands in an audit log.
AI Front Door
One multi-tenant deploy that gives every business its own connector, white label, under an agency’s brand.
Small Business Agent Bench
Our open benchmark of how well AI agents handle real small-business work: 200 tasks, sandboxed tools, deterministic grading.
Every connector we have built for Claude The AI Connector Directory: every small business connector
Four ways in / Set scope, quoted in the conversation
Pick the rung. Talk first.
Connector Blueprint
For software teams whose own engineers will build the connector.
Your engineers start from a design that already answers what directory reviewers ask.
Connector Build
For one product: yours, inside Claude and ChatGPT.
Your customers do their real work in your product from inside the assistant they already use.
Connector Suite
For companies with several products or systems.
One assistant works across everything you run, with a crew of agents doing the multi-step jobs.
Connector Care
For any connector we built, after it ships.
The connector keeps working while the protocol, the models and the directories keep moving.
Not sure a connector belongs in your product yet? Start with an advisory conversation, or read how a server gets into Claude’s directory and how the ChatGPT review differs. If you are weighing whether to hire at all, the guide to hiring an MCP developer says what to hire for, and our comparison of MCP server development companies reads each firm from its own site. Two of us build them, with an agent team we have documented in public.
Straight answers
What teams ask before they hire.
What is an MCP server?
An MCP server is a program that gives an AI assistant access to a product’s data and actions through the Model Context Protocol, an open standard. The assistant reads the server’s list of tools and calls them for the person using it. Claude calls a remote MCP server a connector.
What does an MCP server development company do?
It designs the tools the assistant will call, builds the server against your API, secures it with sign-in and an approval step on every write, tests it, publishes it, submits it to the directories and keeps it working as the protocol and the models change. That is the work we do, in that order.
Is an MCP server the same thing as a Claude connector or a ChatGPT app?
A Claude connector and a ChatGPT app can run on one remote MCP server. What differs is each directory’s review: the auth, the tool hints, the test accounts and the listing labels. We build the server once and file both submissions.
How much does MCP server development cost?
It is quoted as a set package in the conversation: the scope, the price and the calendar dates agreed in writing before work starts. Changes to what we built are included. We do not publish prices on the site.
Do you need access to our code?
No. We design from your public API documentation and build against your API with the scopes you grant. If a job needs an endpoint you do not have yet, the proposal says so plainly.
Does a directory listing mean Anthropic or OpenAI endorses you?
No. A listing means people can find the connector in the directory, nothing more. We build to every published rule and file the submission, and Anthropic and OpenAI decide what they list.
Should we build it in-house or hire an MCP server development company?
Your engineers can write the code. What you buy from us is a design that has already been through directory review. If your team would rather write it, the Connector Blueprint hands them the design, an eval set and a working prototype, and it is credited in full if you bring the build to us later.
One next step
Send us your API docs.
Tell us your product and where your API reference lives. We come back with the jobs your customers would do in Claude and ChatGPT, the tools we would build for them, and the package that fits. Or call (406) 250-6076.
Claude is a product of Anthropic. Modern Mustard Seed is an independent studio and is not affiliated with or endorsed by Anthropic. ChatGPT is a product of OpenAI, and we are not affiliated with or endorsed by OpenAI.