Make
Connecting Make's MCP server lets your agents run your Make scenarios and manage your Make account: scenarios, executions, connections, webhooks, data stores, folders, keys, teams and organisations. The server offers 67 tools for managing Make, plus one tool for each active on-demand scenario in your account.
Read How MCP Servers Work in Outermind first if you have not connected an MCP server before.
Before you start
- A Make account. Running scenarios works on every Make plan; the tools that manage your account need a paid Make plan. See Billing notes.
- On-demand scenarios for anything you want agents to run. Each active on-demand scenario becomes its own tool.
- A tenant admin in Outermind to run the wizard and grant the tools.
Make supports two ways to connect:
- OAuth sign-in at
https://mcp.make.com, the known server. This is the usual choice. - A static MCP token at your Make zone's own endpoint,
https://<zone>.make.com/mcp, sent as a Bearer token.
Prepare Make
For OAuth sign-in, there is no app to create: Outermind registers itself with Make automatically when you connect. Do plan what you will tick on Make's consent page, though. It pre-selects no permissions, and an admin who clicks Allow without ticking anything gets a connection with no scopes, so agents can do nothing through it. Tick Run your scenarios, plus the scenario and organization permission groups you want agents to have, before you click Allow.
For a static MCP token, create an MCP token in your Make account with the mcp:use scope, plus the management
scopes for whatever you want agents to manage. Treat the token like a password: anyone holding it can do what its
scopes allow. Note your Make zone, the part of your Make address before .make.com.
Connect in Outermind
With OAuth sign-in
- Open the Capabilities hub, click MCP Servers under Tool Management, and click Add MCP server.
- Under Known server, pick Make (mcp.make.com). The endpoint
https://mcp.make.comis filled in, the wizard shows "Sign in and approve. To use a static MCP token instead, choose Other endpoint, enterhttps://<zone>.make.com/mcpfor your zone and paste the token." and the expected sign-in is OAuth sign-in. Optionally edit Purpose, the one-line description agents see beside this server. Click Next. - On Credential, the wizard says "Outermind registers itself with this server automatically when you connect." Click Connect and sign in to Make. On Make's consent page, tick Run your scenarios and the scenario and organization permission groups agents should have, then click Allow. Make ticks none of them for you.
- On Summary, Outermind discovers Make's tools and makes them active. Under Grant tools to agents, pick an agent to choose an access level for each one; to leave a tool out, use the connection page afterwards.
With a static MCP token
- Click Add MCP server, and under Known server choose Other endpoint.
- Type your zone's endpoint,
https://<zone>.make.com/mcp, with<zone>replaced by your Make zone. Click Next. - On Credential, paste the token into Bearer token and click Connect. Outermind stores it encrypted and never shows it again.
- On Summary, Outermind discovers Make's tools and makes them active; grant them to agents as above.
Scenario tools change as your scenarios change
Make lists one tool for each active on-demand scenario, so the tool list changes whenever you add, remove or change scenarios at Make. Outermind's daily re-sync (or Re-sync on the connection page) picks the change up: a removed scenario's tool leaves your agents at once, and a new scenario's tool becomes active for your agents at once too. See Re-sync and staged operations.
What your agents can do
Agents can:
- run your active on-demand scenarios, each through its own tool;
- look up scenario executions;
- manage scenarios, connections, webhooks, data stores, folders and keys;
- work with teams and organisations.
Make waits up to 25 seconds for a scenario run over OAuth sign-in, and up to 40 seconds with a static token. A run that takes longer keeps going at Make: the agent gets the run's execution ID instead of its result, and can look the execution up later.
Read actions start at tier low, and every other action starts at high; see
Risk tiers.
Running a scenario does whatever that scenario does in the apps it connects to, such as sending email or changing
records. The management tools can change or delete scenarios, connections, webhooks, data stores and keys. Keep
all of these at tier high, and exclude the management tools on the connection page unless agents genuinely need
them.
What is not available
This server covers less than Outermind's earlier built-in Make integration did. Agents cannot use:
- Connection credential test. Agents cannot check that one of the app connections your scenarios use, such as a sign-in to another service, still works.
- Webhook queue statistics. Agents cannot check how many incoming requests are waiting in a Make webhook's queue.
- Synchronous runs over 25 s. Over OAuth sign-in, an agent gets a scenario's result in the same call only when the run finishes within 25 seconds, where the earlier integration waited longer. For a longer run, the agent gets the execution ID and looks the result up afterwards.
Billing notes
Calls through Make's MCP server are not charged per call the way Zapier counts tasks. A scenario that an agent runs uses your Make plan like any other run of that scenario. The tools that manage your Make account need a paid Make plan; on a free plan, only the scenario tools work.