A team may only use approved MCP servers with its agents. What do you configure?
Why D is the answer
An MCP allow list
Allow lists restrict which MCP servers can be used. A prompt instruction would not enforce that.
In detail
- An MCP allow list limits agents to approved servers.
- Review the list as new servers are requested.
Use case
An illustrative scenario, not a real incident.
A security team approves two MCP servers. An allow list means agents can use only those, whatever a developer configures.
See it in different ways
flowchart TD
A{"MCP servers policy enabled?"} -->|No| B["No MCP servers run"]
A -->|Yes| C{"Restricted to an approved list?"}
C -->|Yes| D["Only approved servers run"]
C -->|No| E["Any server can run"]mindmap
root((Agent limits))
Identity
Write access to trigger
Repository
One branch
Branch protection
Tools
MCP allow list
Tool permissions
Network
Firewall allowlistflowchart TD
Q{"Is the server on the allow list?"}
Q -->|"Yes"| A0["Agent may use it"]
Q -->|"No"| A1["Blocked"]Key point: An allow list limits agents to approved servers.
Why the other options are wrong
- A. A longer system prompt
- A prompt is guidance, not a control.
- B. A branch naming rule
- A branch rule has nothing to do with MCP servers.
- C. A larger model
- Model size does not restrict servers.