[ LOADING ]
Connect Sailop to Cursor through MCP and .cursorrules. Cursor chat gains access to all Sailop scanning and generation tools, and the rules file trains Cursor to avoid AI patterns from the start.
The CLI is required for the MCP server. Install it globally so Cursor can find it.
Verify with: sailop --version
The .cursorrules file sits in your project root. It tells Cursor to use Sailop tools when generating frontend code and to avoid known AI defaults. This is a preventive layer — it reduces slop before it is even written.
# .cursorrules — Sailop integration
# This file tells Cursor to use Sailop for frontend quality checks.
When generating or modifying frontend code (HTML, CSS, JSX, TSX):
1. After generating code, call the sailop_scan MCP tool to check for AI patterns
2. If the slop score is above 30, call sailop_fix to apply corrections
3. Use sailop_generate to create unique design tokens when starting new projects
4. Avoid these known AI defaults:
- Blue-to-indigo gradients (HSL 220-260)
- Inter as the only font
- Uniform rounded-2xl borders
- Pure 4px-grid spacing (8, 16, 24, 32)
- backdrop-blur on navbars
- transition-all duration-300 on everything
- Three-column feature grids with icon + heading + text
5. Prefer serif or non-default display fonts
6. Use off-grid spacing values (5, 14, 22 instead of 8, 16, 24)
7. Vary border radii by element type (buttons vs cards vs containers)Save as: .cursorrules in your project root
Add the Sailop MCP server to your Cursor configuration. This gives Cursor chat direct access to all Sailop tools — scan, fix, generate, report, check, and compare.
{
"mcpServers": {
"sailop": {
"command": "sailop",
"args": ["mcp", "--stdio"]
}
}
}This adds Sailop to one project. Create the .cursor directory if it does not exist.
Open Cursor Settings, navigate to the MCP section, and add a new server with the commandsailopand argsmcp --stdio. This makes Sailop available in every Cursor project.
This creates the .cursor/mcp.json and .cursorrules files automatically.
Once MCP is configured, Cursor chat can call Sailop tools. You can ask naturally or be specific.
"Scan this file for AI slop"Cursor calls sailop_scan on the active file and shows the score with findings inline.
"Make this landing page more unique"Cursor calls sailop_scan, then sailop_generate, then applies the generated design system to your code.
"Check if my CSS has any AI patterns"Cursor scans all CSS/TSX files in the project and produces a summary report.
"Generate a design system for this project"Cursor calls sailop_generate with your project context and outputs CSS custom properties, font pairings, and a palette.
"Compare this with my last scan"Cursor calls sailop_compare to show improvement or regression since the last scan.
Both editors get the same MCP tools. The difference is in how you interact with the results.
One npm install, one config file. Cursor chat gets superpowers.