Guide
What is a done-for-you MCP connector?
MCP is the standard that lets Claude reach your tools. A done-for-you connector means you describe the tool and the work, and someone else handles the building, scoping, and upkeep.
A done-for-you MCP connector is a custom connector that a service builds, scopes, and maintains for you, instead of you writing it yourself. It gives Claude tools to read and write to your software inside a conversation you start, acting as your own account with your permissions. You describe the tool and the workflow, and the building is handled for you.
Key takeaways
- MCP is the open standard that lets Claude talk to outside tools through named actions.
- Done-for-you means the building, scoping, and maintenance are handled, not just the code.
- The connector acts inside a Claude conversation you start, as your own account.
- Read, create, and update flow in the chat; delete and send stay behind a confirmation.
- It fits mainstream apps and custom or in-house systems that no directory lists.
What does MCP mean in a connector?
MCP, the Model Context Protocol, is the open standard that lets Claude reach outside tools. A connector built on MCP exposes named actions, like search a record or update a field, that Claude can call during a conversation. It is the wiring between Claude and your software, with each action defined and scoped.
Without a connector, Claude only knows what you paste into the chat. With one, it can call real actions in your system, in sequence:
- 1Look something up.
- 2Draft a change.
- 3Write it back.
Each action is a defined tool with its own inputs and permissions, so the connector is a deliberate set of capabilities rather than open-ended access to everything in your account.
What does 'done-for-you' actually cover?
Done-for-you covers the parts most teams do not want to own: mapping your tool's real schema, defining each read and write action, scoping permissions, handling authentication, and maintaining the connector as your software and the protocol change. You describe the workflow you want; the building and upkeep are handled for you.
In practice, done-for-you usually covers:
- Mapping your tool's real schema, including custom fields.
- Defining each read and write action Claude can call.
- Scoping permissions and handling authentication.
- Maintaining the connector as your software and the protocol change.
Building a connector once is one thing. Keeping it working as APIs shift, fields change, and the protocol evolves is the part that quietly becomes a job. Done-for-you means that ongoing maintenance is part of the arrangement, not a surprise later.
It also covers scoping. Deciding which actions Claude gets, which are automatic, and which wait for a confirmation is a design decision made with you, not left as a default.
What can a done-for-you connector do inside Claude?
Inside Claude, a done-for-you connector can search and summarize your records, then create and update them in the same conversation. The reading and the writing that used to mean switching tabs and clicking through screens happen where you already ask Claude questions. Sending and deleting stay gated behind a confirmation.
Read, create, and update flow naturally in a chat you start, so you can ask a question and act on the answer in one place.
Outbound and destructive actions, like sending a message or deleting a record, are separate tools that wait for your approval. You decide which actions are automatic and which always need a human in the loop.
Is a done-for-you connector safe to use?
A done-for-you connector is safe when it is scoped properly. It authenticates as your own account and inherits your permissions, so it never grants more access than you already have. Read and write actions are separate, and delete and send are gated behind a confirmation. Claude only acts inside conversations you start.
The connector does not run on its own or reach into your tools in the background. It does nothing until you open a conversation and ask, and it cannot exceed what your account is allowed to do.
Because high-stakes actions sit behind a confirmation, the destructive or outbound steps always have a person approving them before anything leaves your system.
Destructive and outbound actions stay gated
Deleting a record or sending a message always waits for a confirmation you approve. Those steps never run automatically.
Does it work for a custom or in-house tool?
Yes, a custom or in-house tool that no connector directory lists is exactly what a done-for-you connector is built for. The connector is mapped to your specific system and schema, with scoped read and write access. If the tool has an API or a database, Claude can work with it through a custom connector.
In our June 2026 review of 264 business tools, 77 percent had no Claude connector at all, and none shipped full read-write out of the box.