Setup Guide
Connect Cutline to Cursor, Claude Desktop, Windsurf, and other AI-powered tools using MCP.
MCP Setup Guide
Connect Cutline's AI product management capabilities directly to your IDE or AI assistant using the Model Context Protocol (MCP).
Watch the Demo
What is MCP?
MCP is an open protocol that enables AI assistants to securely access external tools and data sources. With Cutline's MCP integration, you can:
- Generate pre-mortems without leaving your IDE
- Chat with AI personas about your code and features
- Load and edit project wikis
- Create issues from analysis results
Prerequisites
Before you begin, make sure you have:
- Active Cutline Premium subscription - Upgrade here if needed
- Node.js 18+ installed - Download Node.js
- An MCP-compatible client - Cursor, Claude Desktop, Windsurf, or Antigravity
Step 1: Install the Cutline MCP CLI
Install the Cutline MCP client globally from npm:
npm install -g @kylewadegrove/cutline-mcp-cli
Verify the installation:
cutline-mcp --help
You should see:
Usage: cutline-mcp [options] [command]
CLI tool for authenticating with Cutline MCP servers
Options:
-V, --version output the version number
-h, --help display help for command
Commands:
login Authenticate with Cutline
logout Remove stored credentials
status Check authentication status
help [command] display help for command
Step 2: Authenticate
Run the login command to authenticate with your Cutline account:
cutline-mcp login
This will:
- Open your browser to Cutline's authentication page
- Prompt you to authorize MCP access
- Securely store credentials in your system keychain
Verify your authentication:
cutline-mcp status
You should see:
β Authenticated
User: you@example.com
UID: abc123...
Token expires in: 55 minutes
Step 3: Configure Your MCP Client
Choose your IDE and follow the setup instructions below.
Cursor
-
Open Cursor Settings (
Cmd+,on Mac,Ctrl+,on Windows/Linux) -
Navigate to Features β MCP Servers
-
Click "Edit in mcp.json" or open
~/.cursor/mcp.jsondirectly -
Add the Cutline servers:
{
"mcpServers": {
"cutline-premortem": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "premortem"]
},
"cutline-tools": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "tools"]
}
}
}
-
Restart Cursor completely
-
Verify - Open the MCP panel and confirm "cutline-premortem" and "cutline-tools" appear as connected servers
Claude Desktop
-
Open the config file:
macOS:
open ~/Library/Application\ Support/Claude/claude_desktop_config.jsonWindows:
notepad %APPDATA%\Claude\claude_desktop_config.json -
Add Cutline servers:
{
"mcpServers": {
"cutline-premortem": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "premortem"]
},
"cutline-tools": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "tools"]
},
"cutline-output": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "output"]
}
}
}
-
Restart Claude Desktop
-
Verify - Look for the π MCP icon in the chat interface
Windsurf
-
Open Command Palette (
Cmd+Shift+PorCtrl+Shift+P) -
Search for "Preferences: Open Settings (JSON)"
-
Add MCP configuration:
{
"mcp.servers": {
"cutline-premortem": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "premortem"]
},
"cutline-tools": {
"command": "npx",
"args": ["-y", "@kylewadegrove/cutline-mcp", "tools"]
}
}
}
- Reload Windsurf
Antigravity
-
Open Settings β MCP Servers
-
Click "Add Server" β "Custom Server"
-
Configure:
- Name:
cutline-premortem - Command:
npx - Args:
-y @kylewadegrove/cutline-mcp premortem
- Name:
-
Repeat for
cutline-toolsandcutline-outputas needed -
Save and reload
Step 4: Test Your Setup
Once configured, try these commands in your AI assistant:
Generate a Pre-mortem
Generate a premortem for launching a B2B SaaS analytics platform
targeting mid-market companies with a 3-month timeline.
List Your Pre-mortems
Show me my recent premortems
Chat with a Persona
What does my persona Debra think about adding a mobile app?
Load a Project Wiki
Load the wiki for my current project and summarize the key risks
Available MCP Servers
Cutline provides 4 specialized MCP servers:
| Server | Purpose | Key Tools |
|--------|---------|-----------|
| cutline-premortem | Pre-mortem generation & management | premortem.run, premortem.list, premortem.status |
| cutline-tools | Personas, wikis, and AI chat | personas.chat, wiki.load, wiki.save, agent.chat |
| cutline-output | PDF generation and Q&A | premortem.render_pdf, premortem.qa_answer |
| cutline-integrations | External tool connections | integrations.create_issues |
For most users, cutline-premortem and cutline-tools provide the core functionality.
Troubleshooting
"Authentication required" error
Run the login command:
cutline-mcp login
"Premium subscription required" error
MCP access requires an active Premium subscription. Upgrade your account.
MCP servers not appearing
- Verify the CLI is installed:
cutline-mcp --version - Check that paths in your config are correct
- Restart your MCP client completely
- Check the client's logs for error messages
Token expired after 1 hour
Tokens auto-refresh, but if you see this error:
cutline-mcp logout
cutline-mcp login
"Command not found: npx"
Make sure Node.js is installed and in your PATH:
node --version # Should show v18 or higher
npm --version # Should show 8 or higher
Best Practices
Start Simple
Begin with just cutline-premortem and add other servers as needed.
Use Descriptive Prompts
The AI works best with clear, detailed prompts:
β Good:
Generate a premortem for a mobile grocery delivery app targeting
urban millennials, launching in 6 months with a $2M budget
β Vague:
Make a premortem
Check Status Regularly
Before starting work:
cutline-mcp status
Getting Help
- Documentation: docs.thecutline.ai
- Email: support@thecutline.ai
- Feedback: cutline.canny.io
Next Steps
- Tool Reference - Complete API documentation
- Examples - Common workflows and use cases
- Personas Guide - Get the most from AI personas