Cotlet is a hosted MCP server and REST API for the apps that never built an API of their own. You connect those apps once, and your AI agent, whether that is Claude, ChatGPT, Cursor or Codex, signs in as you and does the work inside them. It approves members, reschedules bookings, publishes posts and exports reports, the tasks people and teams still do by hand in a browser tab. If you already work with AI agents and see that clicking through interfaces is wasted time, Cotlet is built for you.
Think about where a working week goes, whether you run a one-person business or an operations team of twenty. A surprising share of it goes to approving community members, rescheduling clients, copying a newsletter draft into a publishing tool or exporting a spreadsheet for finance. None of it is hard. All of it is repetitive, and all of it happens inside apps someone has to open and click through by hand.
That is exactly the kind of work AI agents were supposed to take off our plates. So why are so many people and companies still doing it themselves?
Your agent is smart. It just can't get in.
Tools like Claude, ChatGPT, Cursor and Codex can act in the world through connectors, most often over the Model Context Protocol (MCP), an open standard for plugging tools into an AI agent. When an app has a public API, someone usually builds a connector for it, and your agent can work in that app.
The trouble is that many of the apps businesses actually run on have no public API at all, or only a narrow, gated one. Community platforms, booking systems, course marketplaces, newsletter tools, niche industry software: they were built for humans with a mouse, not for software. Skool, for example, has no public API and no developer docs. Substack's developer API returns public profile data only. Booksy's API is partner-only.
So you ask your agent to "approve this week's new members and send each one a welcome message," and the honest answer is: it can't. Not because it doesn't understand the task, but because it has no way to sign in.
You asked your agent. It hit a login page. So you clicked through it yourself. Again.
What Cotlet does
This is the gap Cotlet closes. Instead of waiting for every vendor to ship an API, you connect the apps you click through by hand once, and from then on your AI agent signs in as you and does the work. For a company, that means the same connector serves every agent and every script on the team, under one key.
In practice it works in three steps:
- Pick your apps. Cotlet's catalog lists close to a hundred tools that lack a usable API, from Skool and Substack to Booksy, Udemy, Teachable and Pirate Ship.
- Sign in once. You log in to each app yourself, in your own browser. Cotlet keeps only the session, never your password.
- Add Cotlet to your agent. Every connected app shows up as a set of tools in one MCP server, so your agent sees them all and can finish the job: open the app, fill in the form, click send.
The same actions are also available as a REST API, with the same key. That matters if you prefer a scheduled script or an n8n workflow to a chat window. Your agent can move Friday's bookings to Monday on request, and a cron job can export your member list every Sunday night.
A look inside: requesting access
Cotlet grants access per account rather than opening the doors to everyone at once, so the first thing you see after signing up is a short request form. It is worth a look, because it shows how the product thinks about your work: tools first, then the job you want done.
The second step asks what your agent should actually do, in your own words. This is a good exercise even if you never sign up. Writing down "approve new members whose answers mention a business and send each one a welcome DM" forces you to see the hour a week it costs you.
Three jobs worth handing over first
If you are wondering where an agent with access to your tools would pay off, for yourself or for a whole team, start with the tasks that are frequent, rule-based and boring:
- Community admin. Approving membership requests, exporting member lists to CSV, replying to comments, sending welcome DMs. On the Skool API and MCP server page you can see exactly which actions Cotlet covers for a Skool community.
- Publishing. Turning your notes into a newsletter draft, scheduling the post and a short Note to go with it, then pulling the stats a week later.
- Client scheduling. Moving bookings, answering client messages and exporting the weekly report from booking systems built for salons, studios and service businesses.
For the more technical crowd, Cotlet also publishes ready-made scripts on GitHub. The skool-api-mcp-server repository has one small Python file per job, standard library only, for things like exporting members or approving requests from a scheduled task.
"Is it safe to let an agent log in as me?"
It is the right question to ask, and the answer depends on how the access is built. Cotlet's approach is worth knowing as a benchmark, whichever tool you end up using:
- No stored passwords. You sign in to each app yourself; only the session is kept, encrypted with AES-256 in an EU vault and isolated to your account.
- The model never sees your login. Your agent calls a tool; the session is used on the server side.
- Hard lines. No CAPTCHA solving and no bypassing multi-factor authentication.
- You stay in control. Actions that reach other people wait for your OK unless you switch that off per tool, and disconnecting an app deletes the session at once.
You also keep using your own account, which means you remain responsible for following each app's terms. That is a fair trade for getting the clicks back.
The bigger shift: from "Does it have an API?" to "Can my agent use it?"
For years, picking business software came with a quiet question: does it integrate with anything? If the answer was no, you accepted the manual work as part of the deal. AI agents change the calculation. The tools you already know and like don't need to be replaced just because their makers never shipped an API. They need a way for your agent to get in, on your terms.
That is the gap Cotlet is built for. If you or your team rely on tools that still keep you clicking, it is worth requesting access, listing the apps you use, and writing down the one job you would hand over first. New accounts get a free 7-day trial with 5,000 credits, and no card is needed. You can follow the product's progress on Cotlet's LinkedIn page.
This article was produced in partnership with Cotlet.
