The Make MCP server lets an AI assistant work with your Make account: list and inspect scenarios, check execution history, run scenarios, and, with the right permissions, create and edit them. It also reaches connections, data stores, webhooks and folders. Because it can change and run automations, it is a powerful connector that deserves careful permissions.
Quick facts
- Provider: Make
- Server address:
https://mcp.make.com(check Make’s documentation for the current address) - Sign-in: Make account, with a permissions screen you must approve
- Works with: Claude and other apps that support remote MCP connectors
- Typical use: monitor automations, debug failed runs, run on-demand scenarios, draft new scenarios
What you can do with it
| Capability | What it means |
|---|---|
| Explore your account | List organisations, teams, scenarios and folders, and see which apps they use. |
| Monitor and debug | List executions, see whether runs succeeded and open detailed step-by-step results. |
| Run | Run an active scenario on demand with input data, or replay a past run. |
| Build and edit | Create or update scenarios from a blueprint, validate it, and switch scenarios on or off. |
| Connections and keys | List connections and keys and check which apps need them. |
| Data stores and webhooks | Read and edit data store records, and inspect webhook queues. |
| App guidance | Look up apps, their modules and documentation to help plan a scenario. |
Some actions are destructive, such as deleting scenarios or data stores, so keep those off limits unless you really want them.
Before you start
- A Make account, and the organisation and team where your scenarios live.
- An AI app that supports connectors, such as Claude.
- A clear rule for what the assistant may do: read only, run, or also edit.
- Scenarios you have already tested, if you plan to run them.
How to connect it
- Open the connectors or integrations area of your AI app. In Claude it is in Settings, and the location differs between web and mobile.
- Find Make in the connector directory and choose Connect, or add the server address above if your app supports custom connectors.
- Sign in to Make and approve access.
- Switch the connector on for the conversation you are using.
- Ask the assistant to list your organisations and teams so it knows which workspace to use.
- Test with a read-only request, such as “List my scenarios and show which are active.”
Example prompts to copy
Fill in the blanks below, or click a highlighted word in the prompt.
List my scenarios in team [team name] and tell me which are active, which are inactive and which stopped because of an error.
Fill in the blanks below, or click a highlighted word in the prompt.
Show the last 10 executions of the scenario "[name]". For any failed run, explain what went wrong and suggest a fix. Do not change anything.
I want a scenario that watches a Google Sheet for new rows, sends the topic to an AI model and creates a draft WordPress post. Plan the modules and the data mapping first, then wait for my approval before creating anything.
Fill in the blanks below, or click a highlighted word in the prompt.
Run the scenario "[name]" once with this input: [input]. Tell me the outcome and what it did. Do not run it more than once.
Fill in the blanks below, or click a highlighted word in the prompt.
Review the data store "[name]" and list any records that look duplicated or incomplete. Do not delete or edit anything.
Troubleshooting
| Problem | What to try |
|---|---|
| Assistant cannot see your scenarios | Ask it to list your organisations and teams, and confirm it is using the right team. |
| “Insufficient rights” or access denied | Your Make role may not allow that action. Check team permissions. |
| Scenario will not run | A scenario must be active to run. Activate it, and check it is not already running. |
| Run succeeded but results look wrong | Open the detailed execution to see each module’s output before editing the scenario. |
| Too many requests error | Wait a moment before trying again. |
Safety and privacy
- Start read-only. Let the assistant look before you let it run or edit anything.
- Scenarios have side effects. Running one can send emails, create records or spend money, so confirm before each run.
- Do not delete without a backup. Export blueprints first, and keep destructive actions off limits.
- Protect credentials. Connections and API keys are sensitive, so never paste them into chat.
- Watch credit usage. Runs and retries use credits on your plan.
Related in the Library
See the Sheet Topics to WordPress Drafts workflow, the Form Submission to CRM Lead workflow and the Make tool profile.