agentgrade

May 22, 2026 ยท View on GitHub

AgentGrade score for agentgrade.com

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.

License

MIT