Setup
1
Create the config file
Create Replace
.cursor/mcp.json in your project root or your global Cursor config directory:YOUR_OCTOLANE_API_KEY with a personal key from Settings → API keys. Requests use the creator’s current workspace access. Every Octolane plan includes MCP.Cursor also supports OAuth for remote MCP servers. To use it, omit the headers object and enter a personal Octolane key in the connection’s authorization page. See Cursor’s MCP documentation and Octolane authentication.2
Restart Cursor
Restart Cursor to pick up the new MCP configuration. The Octolane tools will be available in Cursor’s AI chat.
3
Verify
Open Cursor’s AI chat and ask:You should see the tools advertised by the connected server. Confirm your workspace with
get_current_user when available. Also try asking what your company brain has learned about a recurring sales objection.Why use Octolane in Cursor
Check customer context while building
When you’re working on a feature request from a specific customer, pull up their deal to see the full context without switching to a browser:Log notes without context-switching
Just finished a customer call? Log it from Cursor:Morning pipeline check during standup
Run a quick pipeline check before your team standup:Check Signal between deploys
While waiting for a deploy, check who’s been visiting your site:Quick data lookups
Need a customer’s email or company details while writing code?Example workflows for developers
Pre-standup pipeline review
Pre-standup pipeline review
Customer debugging context
Customer debugging context
Quick deal creation after a call
Quick deal creation after a call
End-of-day deal update
End-of-day deal update
Task and email check
Task and email check
Signal check for prospecting
Signal check for prospecting
Tips
- Use Cursor’s Agent mode for multi-step CRM workflows (pipeline review + deal updates + note creation in one go)
- Keep queries focused - “Show me stale deals” works better than “Give me a full analysis of our entire pipeline health”
- You can reference CRM data in your code comments: ask about a customer’s requirements, then use the context while building their feature