agentrank
v0.1.0
Published
The Page Speed for the Agentic Web - measure how reliably an AI agent can navigate your site
Maintainers
Readme
agentrank
The Page Speed for the Agentic Web
Measure how reliably an AI agent can navigate your website. Get an Agent Visibility Score (0-100) based on 5 signals: Permissions, Structure, Accessibility, Hydration, and Hostility.
🌐 Try it live at agentrank.it
Quick Start
# Run directly with npx
npx agentrank audit https://example.com
# Or install globally
npm install -g agentrank
agentrank audit https://example.comCLI Usage
# Quick scan (~5 seconds)
agentrank audit https://example.com
# Verbose output
agentrank audit https://example.com --verbose
# JSON output (for scripting)
agentrank audit https://example.com --jsonProgrammatic Usage
import { scanUrl, getGrade } from 'agentrank';
const result = await scanUrl({
url: 'https://example.com',
mode: 'quick', // 'quick' (default) or 'deep'
timeout: 30000, // Timeout in ms
});
console.log(`Score: ${result.agentScore}/100`);
console.log(`Grade: ${getGrade(result.agentScore)}`); // A, B, C, D, or F
// Access individual signals
result.signals.permissions.status; // 'pass' | 'warn' | 'fail'
result.signals.structure.score; // 0-100
result.signals.accessibility.recommendations; // string[]Note: Deep mode (
mode: 'deep') requires a self-hosted browser-use engine. See Deep Mode Setup for Docker configuration.
Agent Visibility Score
| Signal | Weight | What It Measures |
|--------|--------|------------------|
| Permissions | 20% | robots.txt analysis |
| Structure | 25% | Semantic HTML density |
| Accessibility | 25% | ARIA labeling & accessibility tree |
| Hydration | 15% | Time-to-Interactive |
| Hostility | 15% | Bot-blockers (CAPTCHA, etc.) |
MCP Server (IDE Integration)
Start a local MCP server for Cursor or Claude Desktop:
agentrank mcp --port 3000curl -X POST http://localhost:3000/mcp \
-H "Content-Type: application/json" \
-d '{"action": "audit", "url": "https://example.com"}'Requirements
- Node.js 22+
- Playwright browsers (auto-installed on first run)
Links
- Website: agentrank.it
- GitHub: github.com/kiarashplusplus/AgentRank.it
- Issues: Report bugs
License
Apache-2.0 © Kiarash Adl
