Make.com’s MCP server exposes the 3000-app Make scenario ecosystem to AI agents — Claude, Cursor, ChatGPT, and any MCP-compatible client can trigger Make scenarios as tools. The right answer is yes for teams that already have Make scenarios and want AI to trigger them, no for teams that need raw API calls without the scenario overhead. The MCP server speaks JSON-RPC 2.0, the auth is OAuth 2.1, and each Make scenario becomes a callable tool.
Table of contents
- What the Make.com MCP server actually exposes
- The OAuth flow — how the agent gets access
- The scenario trigger — what an AI can do via the MCP server
- The 3000-app ecosystem — what the agent can reach
- The Make.com pricing — the 3000-app cost
- The deployment — hosted by Make, no self-host
- How this fits the rest of the stack
- FAQ
What the Make.com MCP server actually exposes
The Make.com MCP server exposes the user’s existing Make scenarios as MCP tools. Each scenario becomes a tool, the scenario’s input parameters become the tool’s input schema, the scenario’s output becomes the tool’s output. The AI agent (Claude, Cursor) calls the tool, Make runs the scenario, the result returns to the agent.
The right answer is the Make MCP server for a team that has already built the scenarios in Make (a marketing automation, a CRM sync, a data pipeline) and wants the AI to trigger them. The wrong answer is the Make MCP server for a team that has no scenarios — the team is paying for the Make platform without using the automation piece.
The OAuth flow — how the agent gets access
The Make MCP server uses OAuth 2.1 with dynamic client registration. The agent’s first call triggers the OAuth flow, the user is redirected to Make to grant access, Make returns an access token, the agent uses the token for subsequent calls. The token has a scope (the user’s Make account) and an expiry (per the OAuth spec).
The right answer for a team that has multiple agents (Claude Desktop, Cursor, ChatGPT) is one OAuth grant per agent. The right answer for a team that wants to share the scenarios across the team is to grant the agent access to the team workspace, not the individual user.
The scenario trigger — what an AI can do via the MCP server
The agent can call any scenario in the user’s Make account, with the scenario’s input parameters as the tool’s arguments. The common pattern: the agent receives a user prompt (‘send this report to the team’), the agent identifies the right scenario (the ‘send report’ scenario in Make), the agent calls the scenario with the right parameters, Make runs the scenario, the agent returns the result to the user.
The gotcha: the agent does not see the scenario’s internal steps — it sees the input and the output. The right answer is to design the scenario’s input to be self-contained (the agent passes the report data, the scenario handles the rest).
The 3000-app ecosystem — what the agent can reach
Make has 3000+ app integrations (Google Workspace, Slack, Notion, Salesforce, Stripe, HubSpot, and most of the long tail). The agent can trigger any scenario that uses these apps, which means the agent can effectively call any API that Make integrates with. The right answer is the Make MCP server for a team that has built the integrations in Make and wants the AI to leverage them.
The gotcha: the agent is two hops away from the API (agent -> Make -> app). The latency is higher than a direct API call, the failure modes are different (Make scenario fails, not the API call), and the debug path goes through Make. The right answer is the Make MCP server for low-frequency, high-leverage actions. The wrong answer is for high-frequency, latency-sensitive calls.
The Make.com pricing — the 3000-app cost
Make’s pricing is operation-based: the team buys a number of operations per month, every scenario run consumes operations based on the number of modules in the scenario. The free tier has 1000 operations/month. The Core plan is $9/month for 10,000 operations. The Pro plan is $16/month for 10,000 operations + more features. The Teams plan is $29/month for 10,000 operations + team features.
The right answer is to model the operations cost before adding the MCP server — the team’s scenario might consume 50 operations per run, the team’s AI might trigger 100 runs per day, the team’s monthly operations cost is 150,000. The right answer is to design the scenarios to be efficient and to monitor the consumption.
The deployment — hosted by Make, no self-host
The Make MCP server is hosted by Make, not self-hosted. The team’s agent connects to Make’s URL, the team’s scenarios run on Make’s infrastructure, the team pays Make for the operations. The right answer is the hosted server for a team that does not want to operate an MCP server themselves. The wrong answer is the hosted server for a team with strict data residency requirements.
How this fits the rest of the stack
The infrastructure question is a small piece of a larger pattern: the team’s runtime, storage, database, secret store, logs, and deployment platform are all parts of the same platform. The right answer is to model the full stack before the project ships, not after. The RunxBuild hosting calculator is the right place to do that exercise — pick the runtime, the memory tier, the storage, the secret store, and the egress, and the calculator shows what the deploy actually costs at the team’s actual usage.
Useful related references:
FAQ
What is the Make.com MCP server?
A hosted JSON-RPC 2.0 service that exposes the user’s existing Make scenarios as MCP tools. Each scenario becomes a callable tool.
What can the Make.com MCP server do?
It can trigger any scenario in the user’s Make account, with the scenario’s input parameters as the tool’s arguments.
How does the auth work for the Make.com MCP server?
OAuth 2.1 with dynamic client registration. The agent’s first call triggers the OAuth flow, the user grants access, Make returns an access token.
How much does the Make.com MCP server cost?
The MCP server is included in all Make plans. The cost is the operations consumed by the scenarios the MCP server triggers.
Can I self-host the Make.com MCP server?
No — the Make MCP server is hosted by Make, not self-hosted. The right answer for a self-hosted MCP server is to build one or use a self-hosted automation platform with an MCP server.
What is the difference between the Make MCP server and the Zapier MCP server?
Both expose the platform’s automation as MCP tools. Make exposes scenarios; Zapier exposes Zaps. The right answer is the one that matches the team’s existing automation.
Is the Make MCP server free?
The MCP server is included in all Make plans, including the free tier. The cost is the operations consumed by the scenarios.
How do I get started with the Make MCP server?
Create a Make account, build a scenario, enable the MCP server in the Make settings, configure the agent with the Make MCP server URL and the OAuth flow.