agentgrade

Scan any website for AI agent readiness.

Quick Start

npx agentgrade-cli https://example.com

No install needed. Runs directly with npx.

What It Checks

• Payment protocols -- x402, MPP, SPT (Stripe), L402 (Lightning)
• Bazaar discovery -- x402 service catalogs, discoverable endpoints
• MCP servers -- Model Context Protocol endpoints, tools, transport
• OpenAPI specs -- paths, reachability, verification
• AI/Claude plugins -- manifests, validation
• Skills -- skill manifests, endpoint reachability
• llms.txt -- LLM-friendly site descriptions
• agents.txt -- agent permission declarations
• robots.txt -- AI agent-specific directives
• Content negotiation -- agent UA, Accept header handling
• Homepage & meta -- JSON-LD, OpenGraph, canonical URLs
• Infrastructure -- HTTPS, sitemap, cache headers, structured errors
• Identity -- WebFinger, DID, Nostr NIP-05, AT Protocol

Usage

Scan a site:

npx agentgrade-cli https://example.com

Get raw JSON output:

npx agentgrade-cli https://example.com --json

Output

The CLI displays:

• Score -- percentage with per-group breakdown and individual checks
• Payment Rails -- confirmed (live 402) and documented protocols
• Capabilities -- detected agent endpoints (MCP, OpenAPI, plugins, etc.)
• Bazaar -- discoverable paid services
• Report link -- full web report on agentgrade.com

How It Works

The CLI calls the agentgrade.com API, which runs 80+ checks against the target site. Results are displayed in the terminal with the same scoring used on the web scanner.

Links

• agentgrade.com -- web scanner with visual reports
• Knowledge Base -- guides for improving your score

License

MIT