What is agent readiness?

Agent readiness is how well a website serves AI agents — large-language-model crawlers like GPTBot, ClaudeBot, and PerplexityBot, plus autonomous assistants that take action on the site. A site is agent-ready when an agent can discover what the site contains without rendering JavaScript, retrieve content in machine-readable form, respect access-control intent, act on exposed capabilities (API, MCP server), transact where commerce is the intent, and trust the content because the site signals editorial quality.

What does the StartupHub.ai Agent Readiness scanner check?

The scanner runs 50+ checks across six dimensions: Discoverability (robots.txt, sitemap, llms.txt, Link headers), Content (HTML noscript, markdown twin, Schema.org, OpenGraph, lang attribute), Access (bot UA negotiation, Accept header, AI policy file), Capabilities (MCP server card, OpenAPI document, structured actions), Commerce (pricing page, structured offers, checkout API), and Quality (author markup, HTTPS, citation, alt text). Each check returns pass / fail / warn / n/a and the site gets a 0–100 score plus a letter grade A+ through F.

Is the Agent Readiness scanner free?

Yes. Free, no signup, no credit card. Available as a web UI at /agent-readiness, a REST API at /api/agent-readiness/scan, a GET one-liner for quick checks, and an MCP server for Claude Desktop, Cursor, and Windsurf. Rate-limited at 10 scans per 10 minutes per IP on the public surface; higher limits available via API key.

When did StartupHub.ai launch the Agent Readiness scanner?

StartupHub.ai shipped the first public version on April 20, 2026 — the original agent readiness scanner. The companion specification (the Agent-Ready Web Standard v1.0) was published May 14, 2026 at startuphub.ai/spec. Several narrower scanners launched in May 2026 covering single specs; ours remains the broadest audit, with 50+ checks across six independent dimensions vs. the 9–22 checks of single-spec validators.

How does the Agent Readiness scanner score sites?

Each of the 50+ checks returns pass / fail / warn / n/a. The per-dimension score is the percentage of applicable checks passed within that dimension. The overall score is the simple average of the six dimension scores, mapped to a letter grade: A+ (90+), A (80–89), B (70–79), C (60–69), D (50–59), F (under 50). The full scoring methodology is published at /spec under CC BY 4.0.

Can I integrate the Agent Readiness API into my own product?

Yes. POST any URL to /api/agent-readiness/scan and get JSON back with total_score, grade, per-category breakdown, individual check results, and a markdown summary. MCP-callable from Claude Desktop, Cursor, and Windsurf via the StartupHub.ai MCP server. Native nodes in Zapier, Make, n8n, and Clay. Free tier covers most use cases; higher volumes via API key. Full docs at /docs/agent-readiness.

What is the difference between the Agent Readiness score and the AEO score?

The Agent Readiness score is a six-dimension audit covering 50+ signals across discoverability, content, access, capabilities, commerce, and quality. AEO ("Answer Engine Optimization") is a marketing term used by some single-spec validators that focus on a narrow slice of the picture, like dual-format markdown serving via HTTP content negotiation. A site can score 100/125 on a narrow AEO check while failing 5 of 6 broader agent-readiness dimensions. The published spec at /spec covers all six.

Agent Readiness · free scanner

Is your site ready for AI agents to use it?

Agent Readiness is a free open scanner for websites, REST APIs, and MCP endpoints. Audits 7 categories — discoverability, content, access, capabilities, commerce (x402 + MPP), and quality. Public leaderboard, paste-ready fixes for Claude and Cursor.

Comparing tools? See how it stacks up against Cloudflare's scanner.

85+ sites scanned today

POST /api/agent-readiness/scan

{
  "url": "anthropic.com",
  "total_score": 84,
  "grade": "A",
  "categories": {
    "discoverability": 94,
    "content":         88,
    "access_control":  82,
    "capabilities":    72,
    "commerce":        80,
    "quality":         86
  },
  "wins": ["llms.txt present", "MCP discoverable"],
  "weaknesses": ["robots.txt missing AI bots"],
  "duration_ms": 1840
}

startuphub.ai/agent-readiness

anthropic.comScanned

84A

Grade A

Agent-ready · 7/7 categories pass

llms.txt · MCP discoverable · 1.84 s

Discoverability94

Content88

Capabilities72

Paste URL

2 sec scan

POST /api/agent-readiness/scan

{
  "url": "anthropic.com",
  "total_score": 84,
  "grade": "A",
  "categories": {
    "discoverability": 94,
    "content":         88,
    "access_control":  82,
    "capabilities":    72,
    "commerce":        80,
    "quality":         86
  },
  "wins": ["llms.txt present", "MCP discoverable"],
  "weaknesses": ["robots.txt missing AI bots"],
  "duration_ms": 1840
}

startuphub.ai/agent-readiness

anthropic.comScanned

84A

Grade A

Agent-ready · 7/7 categories pass

llms.txt · MCP discoverable · 1.84 s

Discoverability94

Content88

Capabilities72

Paste URL

2 sec scan

Don't log my results — score is shown to me but never persisted, never appears on the public leaderboard.

Loading category snapshots…

One scan call. Every signal an agent needs.

Hit the v1 scan endpoint with a URL, get back a category-by-category breakdown plus the raw checks that fired. Wire it into your CI, a competitor-monitor cron, or a Claude agent over MCP and the agent reasons over the JSON directly.

Free tier covers 100 scans / month with full results. Pro and Enterprise unlock unlimited + webhook callbacks when a tracked domain drops a grade.

POST /api/v1/agent-readiness/scan returns score, grade, 6 category scores, and per-check diagnostics
MCP tool scan_agent_readiness for Claude / Cursor / Windsurf — drop-in, no glue code
Webhook callbacks when a domain you track drops below a threshold
Bearer auth, JSON only, cached at the edge for

Is your site ready for AI agents to use it?

Top 100 sites by agent readiness

One scan call. Every signal an agent needs.