How to Use the AI-Readiness Grader to Find Your Biggest Citation Blocker | Fastlook
By Abhijay Tondak, Founder & CEO · Updated August 10, 2026 · 7 min read
The AI-Readiness Grader at fastlook.co/agent-ready fetches your site and scores it from 0 to 100 across four areas that decide whether AI engines can find, read and act on your content. Run it on a URL, read which of the 14 checks fail, and fix the highest-impact failures first. The most common blocker is crawler access.
Key takeaways
- The grader returns a 0 to 100 score and a Level from 0 (Not Ready) to 3 (Agent-Ready) in about a minute, with no sign-up.
- It runs 14 equal-weight, pass-or-fail checks across four areas: Discoverability, Content, Bot Access Control, and API, Auth, MCP & Skills.
- The score is simply the share of checks you pass, so 7 of 14 is roughly 50.
- The most common thing that lowers a score is crawler access: if AI bots cannot fetch you, nothing else counts.
- Fix the failing checks in impact order, then re-grade to confirm the score moved.
What the AI-Readiness Grader measures
The grader answers one question: can an AI agent or answer engine discover, read and act on your site? It fetches your page plus your robots.txt, sitemap.xml, llms.txt and the .well-known files, then scores what it finds.
The result is a number from 0 to 100 and a readiness Level. A score of 0 to 24 is Level 0 (Not Ready), 25 to 49 is Level 1 (Basic Web Presence), 50 to 74 is Level 2 (Emerging), and 75 to 100 is Level 3 (Agent-Ready). The number is the share of checks you pass, so it is easy to reason about: pass more checks, move up a Level.
The four areas it checks
The 14 scored checks are grouped into four areas, each answering a different part of the question. Four bonus signals (llms.txt, JSON-LD structured data, a sitemap link in robots.txt, and Web Bot Auth) are shown in the report but do not change the score.
- Discoverability: can agents find you? robots.txt, a valid XML sitemap, RFC 8288 Link headers, and DNS for AI Discovery.
- Content: can agents read you? Markdown content negotiation, so a page returns clean Markdown when a client asks for it.
- Bot Access Control: do you govern AI access? Explicit AI-bot rules and Content-Signal directives in robots.txt.
- API, Auth, MCP & Skills: can agents act with you? An API catalog, OAuth metadata, auth.md, an MCP server card, an Agent Skills index, and WebMCP.
Put this into practice
See how your site performs across AI engines with a free visibility audit — takes 2 minutes, no credit card.
Run your free auditHow to run it and read your result
Go to fastlook.co/agent-ready, paste a URL, and read three things: the headline score and Level, the per-check pass or fail list grouped by area, and the prioritised fix list at the top.
Here is a real run to show what the output looks like. We graded salaryse.com and it scored 29, which is Level 1 (Basic Web Presence). It passed the fundamentals: the page was reachable, robots.txt was valid, the sitemap was discoverable, and wildcard robots rules covered AI bots. It failed DNS for AI Discovery, Markdown content negotiation, Content-Signal directives, and all six of the API, Auth, MCP and Skills checks.
That pattern is common for a normal marketing site: crawlable and indexable, but with no agent-protocol layer at all. It tells you exactly where the next work is, which is the point of grading before you build.
The most common blocker is crawler access
Before you touch schema or the agent-protocol files, check the two things that gate everything else. First, does your robots.txt allow AI crawlers such as GPTBot, ClaudeBot and PerplexityBot? A blanket Disallow keeps you out of AI answers no matter how good your content is. Second, does your content render on the server, or only after JavaScript runs? Many AI crawlers do not execute JavaScript, so a client-only page reads as empty.
If either fails, fix it first. A site that AI bots cannot fetch or read scores low across the board, and every other improvement is wasted until this is solved.
Fix the failing checks in impact order
Work top down. The early wins are cheap and high-impact; the agent-protocol files are the last mile that separates a good score from Agent-Ready.
- Allow AI crawlers in robots.txt and confirm the page is server-rendered.
- Serve a valid XML sitemap and reference it from robots.txt.
- Restructure the page answer-first, with a self-contained answer directly under each heading.
- Publish an llms.txt manifest and JSON-LD structured data.
- Add the agent-protocol files last: auth.md, an MCP server card, and an Agent Skills index.
When to re-grade
Re-run the grader after each fix to confirm the score actually moved, and on a regular cadence as your site changes. The grader caches a result for about 24 hours per domain, so a same-day re-run may show the cached score until the window rolls over.
Treat the score as a baseline to beat, not a vanity metric. The useful output is the failing-check list, because each failure is a specific, buildable task.
Frequently asked questions
What does the AI-Readiness score measure?
It measures how easily AI agents and answer engines can discover, read and act on your site. The grader fetches your page plus robots.txt, sitemap, llms.txt and the .well-known files, runs 14 equal-weight checks across four areas, and returns a 0 to 100 score with a Level from 0 to 3.
Is the grader free?
Yes. The AI-Readiness Grader at fastlook.co/agent-ready is free and needs no account. You paste a URL and get a scored report in about a minute, including which checks pass or fail and a prioritised fix list. You only sign up if you want Fastlook to implement the fixes for you.
What is the most common thing that lowers the score?
Crawler access. If your robots.txt blocks AI crawlers like GPTBot or ClaudeBot, or your content only renders after JavaScript runs, AI engines cannot read you and the other checks stop mattering. Fix crawler access and server-rendered content before anything else.
What counts as a good AI-Readiness score?
A score of 75 or above is Level 3 (Agent-Ready): agents can find, read and act on your site. 50 to 74 (Emerging) means the basics are in place with gaps. Below 50 means agent-specific signals are largely missing. Many normal marketing sites start in the 25 to 49 band.
How is the grader score calculated?
It runs 14 equal-weight, pass-or-fail checks and scores the share you pass, so 7 of 14 is about 50. Four bonus signals (llms.txt, JSON-LD, a sitemap link in robots.txt, and Web Bot Auth) appear in the report but do not change the number. The score maps to a Level from 0 to 3.
Put this into practice — free.
Get your free AI-visibility audit and see where engines find you today.
Keep reading
Is your site agent-ready?
Most sites score under 30. Check yours in seconds — get a 0–100 agent-readiness score and a prioritized fix list.