Cursor Integration Guide

Overview

Cursor is an AI-native editor with high agentic activity. Cup’n’String applies IDE-aware policy, endpoint governance, secret shielding, and local-service controls.

Support level

Native Integration + Active Proxy & Shielding

What Cup’n’String detects

• Cursor processes
• Workspace activity
• Outbound model/provider endpoints
• Local terminal and agent subprocess patterns where visible

What it governs

• AI provider traffic
• Local service and container access
• Sensitive file exposure
• Policy attribution for agent-driven activity

Recommended policies

• Permit approved providers
• Block unapproved model gateways
• Require approval for local service exposure
• Audit agent-initiated outbound traffic

Setup outline

1. Ensure the Cup’n’String agent is active on the workstation.
2. Open Cursor; the agent identifies the workstation workspace.
3. Apply Cursor-specific provider policies in the Cup’n’String dashboard.

Verification

Submit an LLM request from Cursor and verify that it matches the configured policy actions (allow/warn/block) in your local client.

Troubleshooting

If agent activity fails to attribute, verify if Cursor is configured to use custom system certificates or bypasses standard system proxy configuration.

Known limitations

Some internal hosted workflows may expose less local detail than local agent/proxy workflows.