Yes. You can scan any public URL from the homepage with no sign-up and no credit card, and results come back in seconds. A Pro plan ($19/month) adds the REST API, the authenticated MCP server, and CI/CD gating for teams that want to automate scans — but the core scanner and every fix explanation are free.

How is Agent Ready different from an SEO tool like Lighthouse or Screaming Frog?

General SEO crawlers are built to optimise pages for human search engines — page speed, meta tags, broken links, Core Web Vitals. They do not check the surfaces AI agents actually read: llms.txt, AGENTS.md, markdown mirrors, content negotiation, MCP server cards, A2A agent cards, agents.json, or agent-permissions.json. Agent Ready was built specifically for AI-agent readability and checks all of those in one pass. Lighthouse is a partial exception since its Agentic Browsing category shipped in Chrome M150: it adds six agent-focused audits covering WebMCP, the accessibility tree, layout stability, and llms.txt, and it complements rather than duplicates Agent Ready — see the full comparison at /agent-ready-vs-lighthouse-agentic-browsing. Agent Ready also grades the accessibility tree and layout stability as a separate accessibility sub-score, so that slice is covered too.

Can't I just check these files manually?

You can, and the specs are public — but a manual audit means reading the Vercel Agent Readability Spec, the llmstxt.org standard, SEP-2127, the A2A spec, and the agents.json spec, then hand-verifying 72 conditions across every page and well-known path, and re-doing it on every deploy. Agent Ready runs all 72 checks in seconds and hands back a plain-English fix for each failure, so the audit becomes a remediation list instead of a research project.

Does Agent Ready only check llms.txt?

No. A single-spec llms.txt validator checks one file against the llmstxt.org format. Agent Ready includes those 10 llms.txt checks, but adds 40 site and page checks built on the Vercel Agent Readability Spec plus llms.txt v2 discovery (site-wide and per-page) and 20 checks against agent-protocol manifests — MCP, A2A, agents.json, agent-permissions.json, UCP, x402, NLWeb, API Catalog, Web Bot Auth, Agent Skills Discovery, agent-driven UI (A2UI), MPP (Machine Payments Protocol), AP2 (Agent Payments Protocol), and ACP (Agentic Commerce Protocol) — plus 2 content-integrity checks (cloaking, and robots-vs-manifest coherence), for 72 checks total. On top of that it runs a separate suite of 23 WCAG 2.2 and layout-stability checks over the homepage, reported as its own accessibility sub-score rather than folded into the 71.

What does Agent Ready check that other tools don't?

The full agent-protocol manifest layer, with attribution. Agent Ready validates MCP server cards (SEP-2127), A2A agent cards, Wildcard agents.json, agent-permissions.json, UCP, x402, NLWeb /ask endpoints, API Catalog, Web Bot Auth, and Agent Skills Discovery — alongside llms.txt and the full Vercel Agent Readability Spec — in one pass, and maps every check to the published spec it implements on the /specs page. A2A agent cards, agents.json, NLWeb /ask, and structural AGENTS.md validation remain rare among readiness scanners.

How is Agent Ready different from is-agentic.com or isitagentready.com?

Both are free readiness scanners from platform vendors — is-agentic.com is Vercel's, launched August 2026 and scored by Ora's ranking methodology; isitagentready.com is Cloudflare's — and both are worth running. The structural difference is independence: each platform scanner grades against a methodology its maker authors, while Agent Ready is not owned by a hosting platform, CDN, or model vendor, and grades only against published specifications it didn't write — including vendor-stewarded specs like the Vercel Agent Readability Spec and Web Bot Auth. Every one of the 72 checks is mapped to the spec it implements on the /specs page, each returns a deterministic pass/fail with a plain-English fix, and results are consumable via web, REST API, CLI, MCP server, SDK, and browser extension. See the detailed comparison at /agent-ready-vs-isitagentready.

Agent Ready vs alternatives: agent-readiness tools compared

At a glance

Criterion	Agent Ready	Platform scanners	Manual audit	SEO crawlers	llms.txt validators
Built for	AI-agent readability	AI-agent readiness scoring	Whatever rubric you bring	Human search-engine SEO	llms.txt format only
Grades against	21 published specs, each attributed	Maker's own methodology	The specs you read	SEO best practice	llmstxt.org
llms.txt / llms-full.txt	✓ 10 checks vs llmstxt.org	✓ Checked (depth varies)	By hand	✗ Usually ignored	✓ Its only job
Vercel Agent Readability Spec	✓ 40 site + per-page checks	Own methodology instead	By hand	✗	✗
MCP / A2A / agents.json manifests	✓ All validated in one pass	MCP & commerce; others vary	Possible, slow & error-prone	✗ Not checked	✗ Not checked
Markdown mirrors & content negotiation	✓ Checked per page	Varies by tool	By hand	✗	✗
Total checks	72	Varies by tool	—	SEO-focused set	~6–10 (one file)
Fix guidance	✓ Deterministic fix per check	AI-generated tips & fix prompts	You write it	Generic SEO tips	Format errors only
Output surfaces	Web, JSON API, Markdown, MCP, CLI, extension	Web; CLI/API/MCP on some	—	Web / CSV	Web
Time to result	Seconds	Seconds–minutes	Hours per site	Minutes	Seconds (one file)
Price	Free, no sign-up	Free	Your time	Freemium → paid	Free

What are the alternatives to Agent Ready?

Four approaches cover parts of what Agent Ready does:

Platform readiness scanners. Vercel’s is-agentic.com (launched August 2026, scored by Ora’s ranking methodology) and Cloudflare’s isitagentready.com both score any public URL for agent readiness in seconds, free. Both are worth running — and each grades against a methodology its maker authors, so pair them with a scan graded against the published specs themselves. See how Agent Ready compares to isitagentready.com.
Manual / DIY audits. Read the Vercel Agent Readability Spec, the llmstxt.org standard, and the MCP/A2A/agents.json specs, then hand-check every condition. Total control, but it doesn’t scale and goes stale the moment you ship.
General SEO crawlers. Screaming Frog and similar audit pages for human search engines — meta tags, performance, links — and never look at llms.txt, markdown mirrors, or agent-protocol manifests. Lighthouse is now a partial exception: its Agentic Browsing category adds six agent-focused audits covering WebMCP, the accessibility tree, layout stability, and llms.txt. See how it compares to Agent Ready.
Single-spec llms.txt validators. Lint one file against the llmstxt.org format. Handy, but blind to the other 62 conditions that decide whether an agent can discover, parse, and act on your site.

Why choose Agent Ready?

Independent and attribution-first. Agent Ready isn’t owned by a hosting platform, CDN, or model vendor, and it doesn’t author the rubric it grades you against: every check implements a published specification — including vendor-stewarded specs like the Vercel Agent Readability Spec and Web Bot Auth — and /specs maps each check to its source. The score means the same thing wherever you host.
Built specifically for AI-agent readability. Not an SEO tool with an “AI” tab bolted on — every check maps to how agents discover, parse, and cite your site.
The full agent-protocol manifest layer in one pass. MCP server cards, A2A agent cards, agents.json, agent-permissions.json, UCP, x402, NLWeb, API Catalog, Web Bot Auth, and Agent Skills Discovery — checked together with llms.txt and the Vercel spec.
A remediation plan, not just a score. Every one of the 72 checks ships a clear, actionable fix when it fails.
Free, instant, and everywhere your workflow is. Scan with no sign-up in seconds, and consume results via the web UI, JSON API, Markdown mirrors, an MCP server, a CLI, and a browser extension.
Spec-aligned and evidence-based. Mapped to the Vercel Agent Readability Spec and the llmstxt.org standard, and informed by published research like Princeton’s GEO study on what gets cited in AI answers.

How is Agent Ready different from an SEO scanner?

An SEO scanner optimises for the human-search funnel: rank on a results page, win the click. Agent Ready optimises for the agent funnel: get discovered, parsed, and cited by ChatGPT, Claude, Perplexity, Gemini, and coding assistants — where, per a July 2025 Pew Research study, users click a source link only about 8% of the time. The signals that win there — llms.txt, AGENTS.md, JSON-LD, markdown mirrors, content negotiation, and protocol manifests — are exactly the ones a traditional SEO crawler doesn’t measure.

When is a manual audit or single-spec validator enough?

If you only publish an llms.txt file and want to confirm it parses, a free llms.txt linter does the job. If you’re a spec author who enjoys reading RFCs and you ship rarely, a manual audit is fine. For everyone else — anyone shipping regularly, running more than one page, or exposing MCP/A2A/agents.json — a single scan that covers all 72 conditions and tells you exactly what to fix saves hours and catches the protocol gaps the narrow tools miss. The platform scanners are free too, and a second opinion never hurts — Agent Ready’s job is to be the independent, spec-attributed baseline you automate against.

Frequently asked questions

Is Agent Ready free?: Yes. You can scan any public URL from the homepage with no sign-up and no credit card, and results come back in seconds. A Pro plan ($19/month) adds the REST API, the authenticated MCP server, and CI/CD gating for teams that want to automate scans — but the core scanner and every fix explanation are free.
How is Agent Ready different from an SEO tool like Lighthouse or Screaming Frog?: General SEO crawlers are built to optimise pages for human search engines — page speed, meta tags, broken links, Core Web Vitals. They do not check the surfaces AI agents actually read: llms.txt, AGENTS.md, markdown mirrors, content negotiation, MCP server cards, A2A agent cards, agents.json, or agent-permissions.json. Agent Ready was built specifically for AI-agent readability and checks all of those in one pass. Lighthouse is a partial exception since its Agentic Browsing category shipped in Chrome M150: it adds six agent-focused audits covering WebMCP, the accessibility tree, layout stability, and llms.txt, and it complements rather than duplicates Agent Ready — see the full comparison at /agent-ready-vs-lighthouse-agentic-browsing. Agent Ready also grades the accessibility tree and layout stability as a separate accessibility sub-score, so that slice is covered too.
Can't I just check these files manually?: You can, and the specs are public — but a manual audit means reading the Vercel Agent Readability Spec, the llmstxt.org standard, SEP-2127, the A2A spec, and the agents.json spec, then hand-verifying 72 conditions across every page and well-known path, and re-doing it on every deploy. Agent Ready runs all 72 checks in seconds and hands back a plain-English fix for each failure, so the audit becomes a remediation list instead of a research project.
Does Agent Ready only check llms.txt?: No. A single-spec llms.txt validator checks one file against the llmstxt.org format. Agent Ready includes those 10 llms.txt checks, but adds 40 site and page checks built on the Vercel Agent Readability Spec plus llms.txt v2 discovery (site-wide and per-page) and 20 checks against agent-protocol manifests — MCP, A2A, agents.json, agent-permissions.json, UCP, x402, NLWeb, API Catalog, Web Bot Auth, Agent Skills Discovery, agent-driven UI (A2UI), MPP (Machine Payments Protocol), AP2 (Agent Payments Protocol), and ACP (Agentic Commerce Protocol) — plus 2 content-integrity checks (cloaking, and robots-vs-manifest coherence), for 72 checks total. On top of that it runs a separate suite of 23 WCAG 2.2 and layout-stability checks over the homepage, reported as its own accessibility sub-score rather than folded into the 71.
What does Agent Ready check that other tools don't?: The full agent-protocol manifest layer, with attribution. Agent Ready validates MCP server cards (SEP-2127), A2A agent cards, Wildcard agents.json, agent-permissions.json, UCP, x402, NLWeb /ask endpoints, API Catalog, Web Bot Auth, and Agent Skills Discovery — alongside llms.txt and the full Vercel Agent Readability Spec — in one pass, and maps every check to the published spec it implements on the /specs page. A2A agent cards, agents.json, NLWeb /ask, and structural AGENTS.md validation remain rare among readiness scanners.
How is Agent Ready different from is-agentic.com or isitagentready.com?: Both are free readiness scanners from platform vendors — is-agentic.com is Vercel's, launched August 2026 and scored by Ora's ranking methodology; isitagentready.com is Cloudflare's — and both are worth running. The structural difference is independence: each platform scanner grades against a methodology its maker authors, while Agent Ready is not owned by a hosting platform, CDN, or model vendor, and grades only against published specifications it didn't write — including vendor-stewarded specs like the Vercel Agent Readability Spec and Web Bot Auth. Every one of the 72 checks is mapped to the spec it implements on the /specs page, each returns a deterministic pass/fail with a plain-English fix, and results are consumable via web, REST API, CLI, MCP server, SDK, and browser extension. See the detailed comparison at /agent-ready-vs-isitagentready.