Skip to main content
MCP (Model Context Protocol) is an open standard that lets AI agents connect to external tools through a single, structured interface. Instead of configuring individual API calls, you connect your agent to an MCP server and it automatically gains access to everything that server provides — multiple tools at once.

​
What is an MCP Integration?

An MCP server exposes a set of tools to any compatible AI agent. When you connect one to Z360, your agent can see and use all the tools it provides — just like built-in actions, but sourced from the external service. Z360 supports two MCP transport protocols:
  • streamable_http — the current standard, used by most modern MCP servers
  • SSE (Server-Sent Events) — an older transport, still supported by some servers
The MCP server’s documentation will tell you which one to use.

​
Before You Start

You will need:
  • A Z360 account on any plan
  • The URL of the MCP server you want to connect
  • Any authentication headers the server requires — usually an API key or Bearer token
  • To know which protocol the server uses — this is documented by the server provider

​
How to Connect an MCP Server

Step 1 — Open Integrations Go to AI Studio in your Z360 sidebar, then click Integrations. Step 2 — Open the Add Integration Menu Click the dropdown arrow next to Add Integration and select Configure MCP. Step 3 — Enter the MCP Server Details Fill in the MCP Name, Server URL, and Protocol (streamable_http or SSE). Step 4 — Add Authentication Headers Click + Add Header and enter any required headers. If the server requires no authentication, leave this empty. Step 5 — Fetch Available Tools Click Fetch Tools. Z360 will connect to the server and retrieve its list of available tools. Step 6 — Save Click Configure MCP to save. The server appears in your Integrations list with an “MCP” badge. Step 7 — Use the Tools in Abilities Open any ability in AI Studio → Abilities, add an Act node, and select the MCP tools you want to use. Related article: Building Abilities in Z360
What your agent can do: Search your Notion workspace, read pages and database entries, and create new pages during a conversation — useful for logging customer interactions or adding records to a database.Free tier: Notion’s free plan includes API access. No credit card required.Setting up Notion
  1. Go to notion.so/my-integrations and click + New integration.
  2. Give it a name and select your workspace.
  3. Under Capabilities, check Read content, Update content, and Insert content.
  4. Click Submit and copy the Internal Integration Secret. This is your token.
Next, share your pages with the integration. Notion only gives integrations access to pages you explicitly connect.
  1. Open the page or database you want Z360 to access.
  2. Click the … menu, go to Add connections, and select your integration.
Repeat steps 5 and 6 for every page or database you want your agent to access.Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: Notion
  • Server URL: https://mcp.notion.com/mcp
  • Protocol: streamable_http
Add one header: Authorization with value Bearer YOUR_NOTION_TOKEN.Click Fetch Tools — you should see tools like search, create_page, retrieve_page, and query_database. Click Configure MCP to save.Using it in an abilityAdd an Act node and select a Notion tool. For example, use create_page to log a new customer record to a database, or search to look up a customer’s past notes.
What your agent can do: Create and read GitHub issues, search repositories, and manage pull requests — letting your team or customers interact with your dev workflow directly through a conversation.Free tier: Fully free for public and private repositories.Setting up GitHub
  1. Log into GitHub and go to Settings → Developer settings → Personal access tokens → Tokens (classic).
  2. Click Generate new token (classic), give it a name, and set an expiration.
  3. Check repo under Scopes to allow access to repositories and issues.
  4. Click Generate token and copy it immediately. GitHub only shows it once.
Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: GitHub
  • Server URL: https://api.githubcopilot.com/mcp/
  • Protocol: streamable_http
Add one header: Authorization with value Bearer YOUR_GITHUB_TOKEN.Click Fetch Tools — you should see tools like create_issue, get_issue, list_issues, and search_repositories. Click Configure MCP to save.Using it in an abilityAdd an Act node and select a GitHub tool. For example, use create_issue to automatically open a bug report when a customer reports a problem through your chat widget.
What your agent can do: Create issues, update statuses, assign tasks, and search projects in Linear — connecting customer conversations directly to your team’s issue tracker.Free tier: Linear’s free plan includes API access. No credit card required.Setting up Linear
  1. Log into Linear and go to Settings → API → Personal API Keys.
  2. Click Create key, give it a label like “Z360”, and copy the key.
If you want the agent to create issues in a specific team, open that team in Linear and note the ID from the URL.Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: Linear
  • Server URL: https://mcp.linear.app/sse
  • Protocol: SSE
Add one header: Authorization with value Bearer YOUR_LINEAR_API_KEY.Click Fetch Tools — you should see tools like create_issue, update_issue, search_issues, and list_teams. Click Configure MCP to save.Using it in an abilityUse Ask nodes to collect issue details, then add an Act node selecting create_issue. Map the title, description, and team to the collected variables. Add a Speak node to confirm the issue was created.
What your agent can do: Create tasks, read project lists, and mark tasks complete in Todoist — capturing action items from conversations automatically.Free tier: Todoist’s free plan includes API access. No credit card required.Setting up Todoist
  1. Log into Todoist and go to Settings → Integrations → API token.
  2. Copy the API token on that page.
If you want tasks added to a specific project, note the project ID from its URL. Otherwise tasks go to your Inbox by default.Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: Todoist
  • Server URL: https://mcp.todoist.com/mcp
  • Protocol: streamable_http
Add one header: Authorization with value Bearer YOUR_TODOIST_TOKEN.Click Fetch Tools — you should see tools like create_task, get_tasks, update_task, and get_projects. Click Configure MCP to save.Using it in an abilityAfter collecting a customer’s request or an action item, add an Act node selecting create_task. Map the task content to a collected variable. Add a Speak node to confirm it was added.
What your agent can do: Search the web in real time and return current information to customers — useful for questions beyond your knowledge base like business hours, pricing, or recent news.Free tier: 2,000 searches per month free. No credit card required.Setting up Brave Search
  1. Go to api.search.brave.com and click Get Started.
  2. Create an account and subscribe to the Free plan (Data for AI tier).
  3. Open API Keys in your dashboard and copy your key.
Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: Brave Search
  • Server URL: https://mcp.search.brave.com/sse
  • Protocol: SSE
Add one header: x-subscription-token with your Brave API key.Click Fetch Tools — you should see tools like web_search and local_search. Click Configure MCP to save.Using it in an abilityAdd an Act node selecting web_search. Pass the customer’s query as the search argument. Follow with a Speak node summarizing the results in plain language.
What your agent can do: Read, create, and update records across your Airtable bases — connecting conversations directly to a database your whole team can see.Free tier: Up to 1,000 records per base. No credit card required.Setting up Airtable
  1. Go to airtable.com and create a free account.
  2. Go to airtable.com/create/tokens and create a new token.
  3. Add scopes data.records:read and data.records:write, select your bases under Access, and click Create token. Copy it immediately.
  4. Open your base and note the Base ID from the URL — it looks like appXXXXXXXXXXXXXX. Also note your exact table names.
Setting up in Z360Go to AI Studio → Integrations, click the dropdown next to Add Integration, and select Configure MCP. Fill in:
  • MCP Name: Airtable
  • Server URL: https://mcp.airtable.com/mcp
  • Protocol: streamable_http
Add one header: Authorization with value Bearer YOUR_AIRTABLE_TOKEN.Click Fetch Tools — you should see tools like list_records, create_record, update_record, and search_records. Click Configure MCP to save.Using it in an abilityUse Ask nodes to collect the customer’s information, then add an Act node selecting create_record. Specify the base ID, table name, and field values. Add a Speak node to confirm the record was saved.

​
Frequently Asked Questions

Custom Integrations connect to one specific API endpoint at a time — you configure each call individually. MCP connects to a server that exposes many tools at once. With MCP, you connect once and immediately get access to everything the server provides.
Use MCP when the service has a published MCP server — it is faster to set up and gives your agent more capabilities automatically. Use a Custom Integration when no MCP server exists for that service, or when you need to call a very specific endpoint.
There is no documented limit. You can connect multiple servers and your agent will have access to tools from all of them simultaneously.
No. You need to add MCP tools to your ability workflows using Act nodes, just like any other integration. The tools only fire when the workflow reaches an Act node that uses them.
streamable_http is the newer, more efficient protocol and is recommended for most servers. SSE is an older method still used by some servers. Check the documentation for the server you are connecting to confirm which one it supports.
Yes. As long as the server is publicly reachable via a URL, Z360 can connect to it. This is useful if you want to build your own server for internal tools or proprietary data.
Z360’s MCP configuration supports static headers. For OAuth-based servers, generate a long-lived Personal Access Token from the service and pass it in the Authorization header. Full OAuth flows are not handled automatically.
No. You connect the full server and all of its tools appear in your Act node list. You simply do not have to use the ones you do not need.
Go to AI Studio → Integrations, find the MCP card, click Edit, update the header value, and click Save. No need to reconnect from scratch.
Not every service has a public MCP server yet. Search for “[Service Name] MCP server” on GitHub — community-built servers exist for many popular tools. If nothing exists, use a Custom Integration to connect directly to the service’s REST API instead.

​
Troubleshooting Guide

Z360 cannot reach the server. Check that the Server URL is correct and includes the full path, the protocol matches what the server expects, your authentication header is correctly formatted, and the server is online.
The server connected but returned an empty tools list. Check that your API token has the necessary permissions and scopes, and that you are connecting to the correct endpoint path for that server.
Check that the arguments passed to the tool match what it expects in name and data type, your token has not expired, and the resource the tool is trying to access actually exists in the third-party service.
Authentication is failing. Check that your token is valid and not revoked, the header key matches exactly what the server expects (check case), the token format is correct, and your token has the correct scopes.
Check that the Act node is selecting the correct MCP tool, all required arguments are being passed from earlier nodes, and the ability was saved after configuring the Act node.
Go to AI Studio → Integrations and confirm the MCP card shows a tool count. Refresh the page and open the ability again. If tools are still missing, click Fetch Tools on the MCP configuration to re-sync.
The authentication token has likely expired or been revoked. Generate a new token from the third-party service and update the header value in your MCP configuration. Also check whether the server URL or protocol has changed.
Related help center articles: