Create an MCP key
An MCP key is a personal API key that lets an assistant or a script, such as OpenAI Agent Builder, Cursor, or Claude Desktop, connect to Performance Hub through its MCP server. You create and manage these keys yourself from User Preferences; each key belongs to your account and inherits your access.
Before you start
- MCP keys are tied to your user account, not to a single facility - the key carries whatever access your account already has.
- The new key's secret value is shown only once, at creation. Have somewhere secure ready to store it (a password manager or your MCP client's settings).
- You'll need two things to configure your MCP client: the Server URL and an active key. Both are on the MCP & API Keys tab. If you have no keys yet, the tab says No MCP keys yet.
Open MCP & API Keys
Open User Preferences

Open the MCP & API Keys tab
Read the callout, then the key list
"All requests made with these keys are logged in the audit logs." "Retention settings are configured by your organisation administrator." View logs opens Audit Logs filtered to your MCP calls.
Your MCP Keys has View MCP Docs, Download CSV and Create Key. With at least one key, MCP Server Configuration shows the Server URL https://app.performancehub.co/mcp and tells you to use any active key from the list.

Create a key
Start a new key
Give it a label

Copy the key now

Security: Treat an MCP key like a password. Don't paste it into shared documents, screenshots, or chat. If a key is ever exposed, deactivate it (and delete it) straight away and create a replacement.
Connect your MCP client
Use the details from the MCP Server Configuration panel to set up your client:
| Field | Where to find it | Notes |
|---|---|---|
| Server URL | MCP Server Configuration, on the MCP & API Keys tab | Click Copy to copy it. This is the address your MCP client connects to. |
| API Key | Any Active key in the list | Use the secret value you copied/downloaded when the key was created. |
For client-specific install snippets (Cursor, Claude Desktop, and more), see Set up & integrate.
Manage existing keys
Each key in the list shows its redacted identifier, a status badge, an editable Label, and its Created (and, if used, Last used) timestamps, alongside lifecycle controls.

Key statuses
| Status | Meaning |
|---|---|
| Active | The key works and can authenticate MCP requests. |
| Deactivated | The key is disabled and will be rejected. It can be reactivated or permanently deleted. |
Available actions
| Action | When it's shown | What it does |
|---|---|---|
| Label (edit field) | Always | Rename a key inline. A green check confirms Changes Saved. |
| Deactivate | Active keys | Disables the key immediately. MCP clients using it will stop working. |
| Reactivate | Deactivated keys | Re-enables a previously disabled key. |
| Delete | Deactivated keys only | Permanently removes the key. A key must be deactivated before it can be deleted, and this cannot be undone. |
| Download CSV | Header, once after first key creation | Exports your key list (identifiers, labels, status, timestamps) as a CSV. |
To retire a key safely: Deactivate it first (this immediately blocks any client still using it), confirm nothing breaks, then Delete it if you no longer need it.
Allow integration secrets
Each key has an Allow integration secrets switch, off unless you turn it on. "Lets scripts using this key read API keys and passwords from External Integrations when they ask for them." Turning it on asks you to confirm. Leave it off unless you trust the script that will use the key. A script with the switch on can read those secrets as you.






