Inocta GEO Status (Chrome/Edge Side Panel Extension)

This extension analyzes the current page you are viewing and shows a GEO dashboard aimed at AI agent visibility (how likely generative engines are to retrieve/cite your content).

It also generates downloadable helper files:

• llms.txt (structured summary)
• markdown.md (best-effort page-to-markdown export)
• humans.txt (generated from the form in the side panel)

How the extension works (simple overview)

1) Side panel UI (sidepanel.html, sidepanel.js, sidepanel.css)

• Opens when you click the extension icon.
• Asks the active tab for an audit using chrome.tabs.sendMessage(...).
• Renders the score, metrics, file status, and suggestions.
• Lets you export a JSON audit and download generated files.
• Includes an in-app Guides popover with three built-in GEO documents (rendered from Markdown).

Header information hierarchy (UI rule)

To keep the side panel easy to scan, the header is organized into three levels:

• Level 1 — Menu: the top navigation tabs (Dashboard / llms.txt / markdown.md / humans.txt).
• Level 2 — Actions: text-only buttons (Guides / Open report / Export / Refresh).
• Level 3 — Signals: the audit context controls (Citation vs Compliance mode + detected intent badge).

2) Page analyzer (content.js)

Runs inside the website tab and returns:

• geo: scores + detailed metrics + suggestions
• fileStatus: whether key site files exist (robots/sitemap/llms/humans)
• robotsInsights: extra context used by the UI (see below)
• files: generated llms.txt, markdown.md, and humans.txt

Intent model (classic user-goal) + commerce context

The tool separates two concepts:

• Intent (user goal): what the user is trying to do on this page
  • INFORMATIONAL: learn/understand
  • NAVIGATIONAL: find a destination/hub (homepage, store locator, support)
  • TRANSACTIONAL: complete an action (buy, checkout, signup, booking)
  • COMMERCIAL: evaluate options before buying (comparisons, “best”, reviews)
• Commerce context (page type): whether the page looks like ecommerce, and what kind
  • page.content_type: ecommerce_homepage, category_page, product_page, search_results_page, cart, checkout, store_locator, unknown
  • page.commerce_intent: evidence signals + a confidence score

This prevents “ecommerce” from becoming a confusing competing intent. Instead, ecommerce is captured as page context while intent stays about user behavior.

How intent affects scoring

• Citation Score uses intent-aware weights. Different intents weight metrics differently.
• Compliance Score does not change by intent; it is a Machine & AI Readiness audit (access + identity + basic extractability). It does not measure content quality or citation likelihood.

Sitemap detection (robots-first)

Some sites do not host their sitemap at /sitemap.xml (they might use sitemapindex.xml, .gz files, or multiple sitemap URLs).

To avoid false “missing sitemap” results, the extension now:

1. Fetches /robots.txt
2. Extracts all lines like:
   • Sitemap: https://example.com/sitemapindex.xml
3. Verifies those URLs look like a sitemap (non-HTML response and sitemap/XML markers).
4. If none are found, it falls back to common paths such as:
   • /sitemap.xml
   • /sitemapindex.xml
   • /sitemap.xml.gz

Verification strategy (fast but robust)

• If robots.txt lists many sitemaps, the extension verifies them with bounded concurrency and stops once it finds the first valid sitemap (to keep audits fast).
• If there are only a few sitemap URLs, it verifies all of them.

In the UI, when the sitemap is verified, the sitemap tile links to the real verified sitemap URL, not a forced https://host/sitemap.xml.

AI crawler blocking warning (warning-only)

Your site can score well on-page but still be hard for AI systems to access if robots.txt blocks AI crawlers.

The side panel shows a warning banner when robots.txt appears to block common AI bots, such as:

• GPTBot
• Google-Extended
• CCBot
• ClaudeBot
• PerplexityBot
• Amazonbot

Important:

• This is informational only (it does not change your GEO score).
• robots.txt policy is a business/legal decision. The extension warns you so you can make an informed choice.

How “blocked” is decided (more accurate)

The extension parses robots.txt into user-agent groups and rules, then evaluates access using common precedence rules:

• Most specific path wins (longest matching rule path)
• If there’s a tie, Allow: beats Disallow:

For the warning banner, we focus on whether each listed AI bot is effectively blocked at the site root (/).

Soft 404 (HTML) vs real files

Some sites return an HTML “404 page” when you request a missing file (example pattern: /404?url=/llms.txt). This is called a soft 404.

The extension labels a file as Soft 404 (HTML) when the response looks like HTML, even if the HTTP status is 200.

To avoid false soft-404s, the extension will not mark a sitemap as soft-404 if the content looks like real XML (e.g., contains <?xml, <urlset, or <sitemapindex) even if the server sends the wrong Content-Type header.

Notes / limitations

• Some websites block extension content scripts on certain pages or require a refresh after installing/updating the extension.
• robots.txt rules can still be complex (wildcards, unusual formatting, vendor-specific behaviors). The warning is designed to be helpful, not a legal guarantee.

Export JSON (two types)

The side panel Export menu currently provides:

1) Raw JSON (tool export)

• A direct export of the audit results and supporting context.
• Includes (when available) robots_insights, such as:
  • AI crawler blocking (aiBotsBlocked, aiBotsBlockedDetails)
  • sitemap discovery counts
  • selected baseOrigin (helps with www vs non-www sites)
• Includes page.content_type, page.commerce_intent, and weights to make intent decisions transparent.

2) Report JSON (includes AI prompt)

• Wraps the Raw JSON export inside a single JSON file that also contains a report_prompt string.
• Purpose: you can paste the JSON into an AI assistant and ask it to generate a client-friendly GEO report without recalculating any scores.