Our Agent Readiness Checklist runs 57 checks on whether AI crawlers and agents can reach, read and use your site, then lists the problems to fix, each with a severity and the fix written out. It is where Discoverability starts in the new Opttab, the Agentic Journey OS for brands. Discoverability is the first of our four pillars: Discoverability, Visibility, Optimization and Agentic Commerce.
This post walks through the checklist screen by screen: the crawler grid, the five check groups, and the fix list your team works from.
Why we check discoverability first
Our line for this pillar is "Be found, read and asked about." Being found comes first for a reason. If a robots.txt rule, a CDN setting or a firewall blocks an AI crawler, that crawler cannot read your pages, and every later step in the agentic journey starts from a hole.
Visibility tracking, content work and shopping work all assume an assistant can reach your site. We built the checklist to test that assumption before you spend on anything built on top of it. It is also the first thing we suggest you run on the new Opttab.
Where to find the Agent Readiness Checklist
Open the Dashboard and click the Issues tab, or click the Discoverability Score card at the top of the Dashboard. Both take you to the same page. The header says which domain we scanned and when, with a Rescan button on the right. Click Rescan after you ship a fix to see it reflected.
The public page for the Agent Readiness Checklist describes the same checks.
The AI crawler grid
The first block answers a basic question: can AI crawlers read your site at all? We check 12 AI crawlers against your robots.txt rules for every page the scan discovered, and show a check or a cross for each.
The grid groups crawlers by company. For OpenAI we check GPTBot, OAI-SearchBot and ChatGPT-User. For Anthropic, ClaudeBot, Claude-SearchBot and Claude-User. For Perplexity, PerplexityBot and Perplexity-User. Then Google-Extended for Google, Applebot-Extended for Apple, CCBot for Common Crawl and Meta-ExternalAgent for Meta. A second row covers the classic search engines: Googlebot, Bingbot and Applebot.
Each tile names the bot and the company behind it, with a check or a cross. The search engine bots sit in their own group below the AI crawlers.
One clarification on Meta. Meta-ExternalAgent appears here only because it is a crawler we check for access. Opttab does not track Meta AI answers.
We split each company into several bots because they do different jobs. A training crawler, a search crawler and a bot fetching a page for a live user request are often governed by separate robots.txt lines. A rule written for one can block another without anyone noticing.
The 57 checks and their six statuses
Below the grid, one summary line counts all 57 checks by status: passed, not passed, to review, to set up, waiting and not assessed. Each group repeats the same breakdown for its own checks, and every check is a card with a title, a one-line plain description, a status icon and a problem count when something fails.
Each card states what it tests in one line. A check mark means it passed, a warning sign needs a look, and a problem count shows how many issues it found.
Not passed is your fix list. We read to review and to set up as items that need a person on your team to look at something or connect a source.
The 57 checks fall into five groups. Below is each group with a few of its checks.
Discoverability
This group asks whether a crawler can find your pages and trust what it finds. Origins, DNS and TLS checks that your domain resolves and serves a valid HTTPS certificate. HTTP response and redirects checks that pages answer with 200, without error codes or long redirect chains. Sitemap and internal-link checks confirm your important pages are listed and reachable from the homepage.
Structured data validity and consistency covers your schema. Our checklist looks at types such as Product, Organization, FAQ and Offer, and at product data such as price, stock, size and returns. Further down, Brand and entity clarity checks that the site states clearly who is behind it. Monitoring connections checks whether Search Console and Analytics are connected. Pages AI answers cite checks that the pages assistants already cite still load and stay open to AI crawlers.
Content accessibility
Once a crawler arrives, can it read the page? Public content and access barriers checks that content is not behind a login, CAPTCHA or bot challenge. Raw HTML versus rendered content compares what a crawler gets without running JavaScript with what a browser shows.
This group also covers Markdown delivery (serving a markdown version when a request asks for text/markdown), the llms.txt index and the llms-full.txt export. llms.txt is helpful but not required, so do not let it hold up the rest of the list. Four checks cover Agent Pages, our agent-readable versions of your pages under Optimize. Agent pages are not cloaking: they carry the same facts as the human page, in a content-negotiated version. Interaction semantics checks that forms and controls are labelled so an agent can use them.
Bot access control
This is where robots.txt gets a closer look. We check that the file can be retrieved and parsed, how well your crawler policy covers the bots that matter, and URL-specific permissions. CDN/WAF and crawler-profile diagnostics look for blocks set at the CDN or firewall rather than in robots.txt.
We also read content-use declarations, such as Content-Signal lines in robots.txt that state whether content may be used for search, AI input or AI training. Web Bot Auth and verified provider crawler visits (from server logs) round out the group.
Protocol discovery and Commerce
The last two groups look past the page to the machine-readable entry points agents use. Protocol discovery checks for an MCP server card at /.well-known/mcp/server-card.json, an A2A Agent Card, published Agent Skills, an API catalog (RFC 9727) and OAuth discovery documents, among others.
Commerce is marked informational. It checks for the agentic payment and checkout protocols: UCP (Universal Commerce Protocol), ACP (Agentic Commerce Protocol), AP2 agent payment authorization, x402 payment challenges over HTTP 402, and MPP (Machine Payments Protocol).
Commerce checks are informational. Each card names the exact profile, file or response it looks for, such as the /.well-known/ucp profile or an HTTP 402 payment challenge.
Read this group as a status report rather than a to-do list. It shows which of these protocols your site already advertises, so you know where you stand before an agent tries to buy from you.
Problems to fix
At the bottom, Problems to fix turns failed checks into a work list. Each row has the problem in plain words with the fix underneath, a Status (Critical, Medium, Low or Info), a Category such as Indexability, Response, Entity clarity or Crawlability, and the detection date.
Every problem comes with the fix written out. Status shows how urgent each one is, and Category tells you which part of the stack it belongs to.
The fixes are specific. A site that serves plain http gets "Redirect every http URL to its https equivalent with a permanent (301) redirect." Missing robots.txt is an Info item: nothing is needed unless you want to exclude paths or list your sitemap.
Opttab finds the issues and suggests fixes, ranked by impact. Your team applies them. Our advice is to route each row to whoever owns that layer on your side: a developer for redirects and the CDN, an SEO lead for structured data, a content owner for a missing about page. Work from Critical down, then click Rescan.
For a page-by-page view, Optimize > Site Audit scores every analyzed page for technical SEO, content quality and answer engine optimization.
Plan notes and next steps
Technical GEO checks are on every plan, from Lite up. Agent pages scale by plan: Lite 10, Starter 20, Growth 100, and Enterprise unlimited (fair use). The pricing page lists the rest. The fuller Shopping Insights readiness audit, which checks product data for AI shopping and agents, is on Growth and Enterprise and is covered in our post on Shopping Insights.
Two follow-ups, both from Starter up, build on a clean result. Website Analytics shows which AI crawlers actually visit, once you connect a crawler log source. Prompt Coverage shows whether you track the questions buyers ask. The launch post on the new Opttab shows how the pillars connect.
Existing customers can open the checklist in the account they already use at app.opttab.com. If you are new, Start your agentic journey or Book a demo.
