MCP Servers

Connect MCP Servers to Glue

Glue AI can now act as an MCP client, allowing it to integrate with applications that expose an MCP server. When an MCP server is connected to Glue, Glue AI can understand your natural language requests and translate them into API-level operations within your connected applications.

If you’re new to MCP: Model Context Protocol is a new open standard that gives language models a uniform way to access external data and tools. This means Glue AI’s capabilities will automatically expand as the MCP ecosystem expands. Learn more about MCP here.

---

🧠 How Glue works with MCP

Glue uses MCP (Model Context Protocol) to connect with third-party applications through a standardized interface.

Key Roles

• Glue AI acts as the MCP client, sending requests.

• The connected application have an MCP server, which receives and responds to those requests.

Types of MCP Servers

To connect an application to Glue, it must have an available MCP server. These come in two forms:

Type	Description	Setup	Examples
Remote Server	Remotely hosted and maintained by a 3rd party app.	Easy	Zapier Linear Sentry
Open Source Community Server	Must be installed and hosted locally on your device.	Difficult	Figma, Notion, GMail, Hubspot - and more can be found here

🔍 Don’t see a server for the app you need?

Some apps may not yet offer MCP, but more are being developed and released regularly.

---

🔧 Step-by-Step Setup Guide

1) Find a compatible MCP server for your app

• Remote / hosted servers – if the app has an official remote MCP server, we recommend using that over a community server. Glue supports the SSE (Server-Sent Events) transport only. Streamable HTTP support will be available soon.

• Community servers – look for one that supports either SSE or stdio transport. You can use a proxy to translate stdio to SSE—see below.

• Helpful directories:

  • MCP Servers Repository - Complete collection of reference implementations and community servers
  • Awesome MCP Servers - Curated list of MCP servers

2) Choose how you will run the MCP server

The setup steps differ based on the type of MCP server you're using — remote or locally hosted community servers with either SSE or stdio transport support.

Scenario	What you need	Glue transport
Remote, hosted server	Just the server’s public URL (from the provider)	SSE ➜ connect directly
Local, community server (SSE)	Run the server + ngrok	SSE ➜ connect via ngrok
Local, community server (stdio)	Run the server + mcp-proxy + ngrok	stdio ➜ mcp-proxy ➜ connet via ngrok

---

☁️ Using a remote/hosted MCP Server (SSE)

These are hosted by the third party application itself. Setup is minimal.

1. Follow the provider's setup instructions.

2. Copy the server’s SSE URL (e.g. https://mcp.linear.app/sse).
3. Skip to Step 3 · Connect the MCP Server to Glue.

🖥️ Using a local community MCP Server (SSE or stdio)

Each community server will have an open source instructions list that needs to be followed and will vary based on the app. These are the general steps that will be required.

1. Install the MCP Server to your local machine

• Follow app-specific instructions (usually in their GitHub README).

• Prerequisites may include tools like Node.js, Homebrew, etc.

# Example: Linear MCP server delivered via npx
# replace with the command shown in the server’s README
npx @ibraheem4/linear-mcp

1. Install & launch server via mcp-proxy (if needed) – converts stdio ⇄ SSE

brew install uv               # light Python runtime used by mcp-proxy
uv tool install git+https://github.com/sparfenyuk/mcp-proxy
# --pass-environment  → passes your shell’s env vars (API keys, etc.) to the server
# --sse-port 9000     → local port on which mcp-proxy will expose an SSE endpoint
# everything after that is the *command* to start the MCP server
mcp-proxy --pass-environment --sse-port=9000 \\
  npx @ibraheem4/linear-mcp

1. Expose your local SSE endpoint to Glue (Glue cannot reach localhost)

• Install a proxy like ngrok if you have not already

• ngrok shows a forwarding URL like https://<random>.ngrok-free.app.

# Install & authenticate ngrok
brew install --cask ngrok

# Forward the SSE port (9000) with basic auth
ngrok http 9000 --basic-auth="YOURUSER:YOURPASS"

1. Form the Glue-compatible URL

• This URL will allow Glue to route requests securely to your MCP server.

<https://YOURUSER>:YOURPASS@<random>.ngrok-free.app/sse

💡 Pro Tip: Choose a unique username/password (must be at least 8 characters). Glue will use this to authenticate securely.

3) Add the MCP Server to Glue

1. Open the Glue app.

2. Click your profile image → Preferences.

3. Navigate to the Labs tab → MCP Servers.

4. Click the Add MCP server button.

5. Name your MCP server.

6. Paste the MCP server URL into the Server URL field.

   Reminder the format is either https://<remote-host>/sse or https://<username>:<password>@<random>.ngrok-free.app/sse

7. Click Add server or follow any additional instructions to sign in and authorize the third party app, as prompted.

✅ Glue AI is now connected to the app via its MCP server. You will be taken to a details screen where you can configure custom rules for the server (see step 4).

See the servers that have already been set up and click "Add MCP server" to add more

Example of setting up the official Linear MCP remote server

If you paste in your Server URL before naming the Server, you will get an error message and need to press the retry button after you input a name.

Note: You cannot edit the MCP server configuration once it’s added, simply remove it and re-add if you want to make changes to the URL.

4) Configure Rules for your MCP Server (optional but recommended)

If you've just set up your server, you will be automatically taken to the server details page where you can set rules.

1. To navigate to the server rules, click on the ellipsis next to the server name under the MCP Servers tab.

2. Select configure rules from the dropdown menu.

3. Input your individual preferences for how Glue AI should interact with the server. Examples include storing IDs, naming conventions, and templates.

4. Click save.

---

💬 What Can You Do With Glue + MCP?

Once connected, Glue AI can

• Make real-time queries to your connected app

• Trigger actions via natural language (e.g., “Create a new issue in Linear”)

• Use the rules you've set for the server to customize how the server is called

Each command requires your confirmation before being executed.

📌 Glue AI prompting tips for MCP

• Mention the name of the MCP server you want to take action with

• Use clear, very specific instructions

  • e.g. "Update the status of Issue ID 12345 in Linear to 'In Progress'.”

  • Note: You may need ID numbers in order to run specific tasks, which you can store in the server's custom rules for future requests.
• Break a complicated request into multiple parts

  • e.g. ask Glue AI first to fetch data, then ask it to take action on the data in a separate request

  • Avoid asking for multiple actions in a single prompt

• Specify the format you want the output to be in, such as a table or list — or better yet, save your preferences in AI Rules!

---

Save time by storing default IDs in server rules

When connecting a tool via an MCP server, you often need to provide IDs that identify where or how an action should happen — for example, which project, board, database, or repository to create something in. By saving these IDs in your server's rules, you can save yourself the step of manually providing IDs each time — Glue AI will do it for you.

Here’s how to discover the necessary IDs:

Read the tool's MCP manifest or docs

Most MCP-exposed tools will document their parameters. Look for:

• Required fields (e.g., project_id, repo_name, board_id)

• Defaultable fields (i.e., things you can prefill with a rule)

• Example payloads

Use the tool in Glue to see what it asks for

Try manually triggering the tool from a thread. If Glue prompts you for a value, that’s a strong hint you can set it ahead of time.

Example: If you’re asked “Which Linear project should this go into?”, that means you could write a rule that injects "project_id": "proj_abc123".

---

⚠️ Support & Limitations

Capability	Supported
OAuth Support	✅
SSE (Server-Sent Events) Transport	✅
Streamable HTTP Transport	❌ Not yet
MCP Client Features	❌ Resources ❌ Prompts ✅ Tools ❌ Sampling ❌ Roots