Skip to content
Help Center
Submit a request Sign in

Tags

Connect AI clients with MCP access

Updated Monday, October 26, 2026 · Marc Staine

MCP & AI Tokens helps connect a supported AI client to Supportbench using an agent-bound authorization flow. The selected identity determines the data and actions the client can access.

Applies to: Supportbench Core. Location: Settings → API → MCP AI Tokens. Access and available controls depend on your role and account capabilities.

Before you begin

  • You need permission to edit this configuration. Ask an administrator if the page is available but its editing controls are not.
  • Review the current configuration and where it is used before making a change. Available choices depend on your account, enabled modules, division access and role.
  • Use a test record or a controlled test account for behavior checks. Do not change production routing or send customer messages just to try an example.

Configure and verify

  1. Open Settings → API → MCP & AI Tokens and choose the supported client setup.
  2. Select the intended active agent and review its access.
  3. Complete the client’s token or OAuth connection using the displayed instructions.
  4. Start with a read-only request for a known, permitted article.
  5. Verify that excluded data remains unavailable. Review and revoke access when the client no longer needs it.

Settings reference

Connection setup

SettingWhat it controlsHow to choose
Provider / ClientSetup guidance for ChatGPT, Codex, Claude or Microsoft 365 Copilot where available.Choose the client you actually use and follow its displayed connection method.
AgentActive agent bound to the access; binding cannot be changed after creation.Use the appropriate role and division scope.
Token or OAuth setupProvider-specific credential or authorization instructions.Copy each value into the field named by the client; an environment-variable-name field expects a variable name, not a token.
Connection and audit detailsIdentifies configured access and available audit evidence.Track the client and owner without recording the secret in shared text.
RevokeRevokes the selected access.Verify replacement access before revoking a connection still in use.

Behavior to know

  • Provider UI and authorization methods can change. Use the current setup instructions shown by Supportbench and the client.
  • A successful connection does not authorize every requested operation; role, tenant and record access still apply.

Recommended use

  • Use terminology that agents and customers already understand. Explain the intended meaning in the description.
  • Keep the initial configuration small enough for people to choose consistently. Add options when they represent a real operational difference.
  • Review existing references before renaming, changing scope, or deleting. Where supported, deactivate an obsolete option before considering permanent deletion.

Worked example

  1. Connect an AI client for knowledge-base assistance with the intended agent.
  2. Ask it to read a known article and return its source link.
  3. Check the connected identity and access before allowing supported write operations.

Troubleshooting

ProblemWhat to check
The item is missing from a picker.Check Active, division access, parent hierarchy, enabled modules, and the user’s role. Reopen the consuming form after saving.
The form rejects a save.Read the displayed validation message. Check required values, text limits and permitted scope; do not change unrelated settings to bypass validation.
An existing record still shows an old value.A configuration change and a record change are different operations. Inspect the record and relevant references before assuming the save failed.
  • API → Create and manage API tokens
  • Organization → Configure agent roles and division access
  • Knowledge Base → Configure preferred terminology for AI and search

Was this article helpful?