Apraksts
See whether AI engines can quote your pages — and fix it with one block.
ChatGPT, Perplexity and Google’s AI answers now answer many of your customers’ questions directly. They quote short, clear passages from sites they can read and trust. Georix shows you, from your own server’s data, whether that is happening for your site, and gives you simple tools to improve it. No coding, no account and no API key needed.
After activation, open Georix Start here. Three steps get you to a first result in a few minutes, and a guided tour shows every section with a real example.
Set up once (no account needed)
- llms.txt, llms-full.txt and ai.txt — generated for you and live within seconds of saving. Nothing to upload.
- Local Business schema — address, phone and opening hours for 100+ business types (Dentist, Law Firm, Restaurant, CA Firm and more).
- AI Crawlers — allow or block GPTBot, ClaudeBot, PerplexityBot, Google-Extended and others, written straight to robots.txt.
- Organization, Article and Breadcrumb schema — added to your pages automatically.
Improve your pages (no account needed)
- Answer Capsules — a worklist showing which pages an AI engine could quote, and a block to publish a short, direct answer where it can’t.
- Query Explorer — type a keyword, tick question words (how, what, cost…), and copy a ready-made regex into Google Search Console. Copy and paste only; nothing is connected.
- Answer Booster and GEO Content Audit — editor panels that flag unsourced statistics, weak openings and missing structure while you write.
- FAQ and HowTo blocks with a niche-aware question suggester.
Measure — fills up over time (no account needed)
- AI Traffic — real visits that ChatGPT, Perplexity, Gemini, Copilot and others send to your pages.
- AI Crawler Cost — what AI bots request from your server, and what they send back.
- Reality Check — an honest check of whether your setup is actually being used, even when the answer is “not yet”.
- Verified Bot Identity — spots fake “GPTBot” traffic with reverse-DNS checks.
- Health and AEO Score — a checklist of anything blocking AI engines, plus a site-wide readiness score.
Optional
- Click Drop Finder (formerly “Lost to AI”) — finds question searches that still show up in Google but are getting fewer clicks than before. This is the only feature that needs a connection: a Google Search Console service-account key. It is off by default, and everything else works without it.
Who it is for
Local service businesses (clinics, law firms, accountants, restaurants, salons), bloggers and publishers, SaaS and product sites, and agencies managing client sites.
Honest by design
Georix only shows numbers it can prove from your own data. For example, Google has confirmed it does not use llms.txt — we still generate it because it is free and may matter later, and Reality Check tells you whether anything actually requests it.
No data leaves your server except where described under Privacy and External services below (DNS lookups for bot verification, and the optional Search Console connection).
Privacy
This plugin does not collect, store, or transmit personal data to the plugin author’s servers. Crawler visit logs (bot name, user agent, IP address, timestamp) are stored only in your WordPress database when logging is enabled. All other processing happens on your own site.
AI Traffic Insights (new in 1.1): this feature records only the AI source (for example “ChatGPT”), the landing page path, a UTC date, and a coarse visit count. It does not store IP addresses, does not store raw user-agent strings, and strips query strings from referrers. This data is stored only in your own site’s database and is never transmitted anywhere. It is preserved if you uninstall unless you explicitly opt in to removal.
Answer Booster (new in 1.1): all content analysis runs locally on your own server. Your content is never sent to any external service, and no analysis results are stored.
AI Crawler Cost Report (new in 1.2): stores only per-day, per-bot aggregate counts (bot name, vendor, request count, an estimated byte figure, and a date). It does not store IP addresses or raw user agents. Bandwidth is an estimate — crawler requests multiplied by a rolling average HTML response size sampled from your own site — and is always labelled as an estimate.
Verified Bot Identity (new in 1.2): to confirm a crawler is genuine, the plugin performs reverse-DNS and forward-DNS lookups on the crawler’s IP address, and compares it to IP ranges bundled locally with the plugin. This runs only on a scheduled task, never during a visitor page load. A raw IP is written to a verification queue row only while the check is pending and is deleted the moment the check completes — only the verdict (verified / unverified) and the resolving organisation are kept. No raw IP is ever shown in the admin UI or included in any CSV export. The wording is always “unverified”, never “fake”.
Answer Capsule Engine (new in 1.3): every Liftability Check runs locally on your own server. Your content is never sent anywhere. The only things stored are the target question you typed and the cached verdict, both in your own database as post meta.
Click Drop Finder, formerly “Lost-to-AI Question Finder” (new in 1.3): opt-in and off by default. When connected it retrieves your own Google Search Console rows (query, page, date, impressions, clicks, average position) and stores them in your own database. No page content, visitor data or personal data is sent to Google — only a date range and the dimensions requested. Requests run on a daily scheduled task or when you click Refresh, never on a visitor page load. Credentials are encrypted at rest where the server supports it, deleted on disconnect and always deleted on uninstall. See External services below for the full disclosure.
External services
This plugin does not send site content or personal data to the plugin author’s servers or to third-party APIs.
Google Search Console API (new in 1.3, opt-in, off by default)
The Click Drop Finder is the only feature in Georix that contacts a third-party service, and it does nothing at all until you explicitly connect it.
- What is sent: a signed request to Google’s OAuth token endpoint (
https://oauth2.googleapis.com/token) to obtain a short-lived access token from the service-account key you pasted, and then queries to the Google Search Console API (https://searchconsole.googleapis.com/) for your own verified property. Each API request body contains only a date range, the dimensions being requested (date, query, page), a row limit and a paging offset. No page content, no visitor data, no personal data and no IP addresses are sent. - When it is sent: only on a daily scheduled task, and when you click “Refresh data”. Never on a visitor page load, and never on an ordinary admin page load. Responses are cached for 12 hours.
- What is stored: your own Search Console rows (query, page, date, impressions, clicks, average position) in a new table in your own database, plus your service-account credentials, encrypted at rest with the server’s sodium or openssl extension where available. If neither extension exists, the screen tells you plainly that the credentials are stored as ordinary text.
- Disconnecting: one click. Stored credentials and the cached access token are deleted immediately, and the scheduled task is removed. On uninstall the credentials are always deleted; the stored Search Console data is preserved unless you opt in to removal.
- Google’s terms apply to this API: Google APIs Terms of Service and the Google Privacy Policy. Service: Google Search Console API, provided by Google LLC.
Answer Capsule Engine (new in 1.3, no external service)
All Liftability Check analysis runs locally on your own server against your own rendered content. Nothing is sent anywhere and no analysis result leaves your database.
Site health checks (your own site only)
When you click “Check File Status” or when the Health panel runs validation, the plugin makes HTTP requests from your WordPress install to your own site’s public URLs (for example /llms.txt and /robots.txt). These self-requests confirm that discovery files are reachable. No data is sent to external servers. The Reality Check “visibility” check (new in 1.2) works the same way — it requests your own homepage only, to count the words a crawler can see.
Bot identity verification (new in 1.2, DNS only, your own database)
To confirm a crawler is genuine, Verified Bot Identity performs standard DNS lookups (reverse DNS on the crawler’s IP, then a forward DNS lookup to confirm it) using your server’s resolver. These are DNS queries, not data transfers — no site content or personal data is sent anywhere. IP ranges used for the check are bundled with the plugin and refreshed with each release; the plugin does not fetch them from any external service. All verification runs on a scheduled task, never on a visitor request.
WPML / Polylang (optional, if installed)
If WPML or Polylang is active, the plugin reads language configuration from those plugins locally to group llms.txt output by locale. The plugin does not make additional outbound requests to WPML or Polylang services beyond what those plugins already do on your site.
Bot documentation links (admin reference only)
The AI Crawlers admin tab lists known crawlers with links to each provider’s public documentation (for example OpenAI GPTBot, Google crawlers, Anthropic, Amazonbot). These are display-only reference URLs in the WordPress admin. The plugin does not connect to those services or transmit your site data to them.
Ekrānuzņēmumi
Bloki
Šis spraudnis nodrošina 5 blokus.
- FAQ Block (AEO Schema)
- Llms Txt Generator Faq Block
- HowTo Block (Schema)
- Llms Txt Generator Howto Block
- Answer Capsule
Uzstādīšana
- In your WordPress admin, go to Plugins Add New, search for Georix, then click Install Now and Activate.
- Georix opens Georix Start here. Follow the three steps:
- Add your organisation name and a 1–2 sentence description, then save.
- Open
yoursite.com/llms.txtto see it live. - Add an Answer Capsule to your most important page.
- Optional: click Take the full tour for a guided walk through every section.
Every Georix screen is in the Georix menu on the left. Each screen has a Take the tour button if you want a refresher.
If /llms.txt returns a 404
Go to Settings Permalinks and click Save Changes, or use the Flush Rewrite Rules button on Georix Setup & llms.txt. Georix needs any permalink structure other than Plain.
BUJ
-
Where does the AI answer summary come from?
-
From your page — if your page gives it something to work with. AI engines do not read your whole site and write a summary from scratch; they lift short, self-contained passages that directly answer the question being asked, and they credit the page they lifted from. If no passage on your page answers the question cleanly, an engine will use a competitor’s page that does. That is the whole game, and it is why Georix v1.3 exists: the Answer Capsule Engine tells you whether your page contains a liftable passage and exactly what is wrong with it, and the Answer Capsule block helps you publish one in your own words. Georix never writes the answer for you.
-
Does the Answer Capsule block give me rich results in Google?
-
No, and we will not claim it does. The block outputs clean semantic HTML and no JSON-LD in version 1.3. Google removed FAQ rich results on 7 May 2026 and deprecated HowTo in 2023. The benefit of the block is a well-formed, extractable answer passage that an AI engine can quote — nothing more, and that is worth doing on its own merits.
-
Do I need a Google Search Console connection to use Georix?
-
No. The Click Drop Finder is off by default and entirely optional; every other feature works with no external service and no API key. If you do connect it, it reads only your own property’s data and you can disconnect at any time.
-
How does the CTR-collapse detector know an AI answer took my click?
-
It does not, and it says so on screen. Google’s API does not report whether an AI Overview appeared. What the detector measures is a specific, checkable shape in your own data: a query whose impressions stayed within 20% across two consecutive 45-day windows while its click-through rate fell by 40% or more. That is a common fingerprint of an answer appearing above you, but a redesigned results page, a new competitor snippet or seasonality can produce the same shape. Treat it as a shortlist worth investigating, not a verdict.
-
Do I need to create the llms.txt file manually?
-
No. The plugin creates the URLs automatically through WordPress. There is no file to upload, no FTP access needed, and no file permissions to configure. Visit
yoursite.com/llms.txtimmediately after saving your settings and it is live. -
My /llms.txt returns a 404 — how do I fix it?
-
Go to Settings > Permalinks and click Save Changes. This flushes WordPress rewrite rules. If the 404 persists, use the Flush Rewrite Rules button on the plugin settings page. Note: the plugin does not work if your permalink structure is set to Plain (default WordPress install). Change it to Post name or any other option.
-
Does this plugin require any API keys?
-
No. Everything works without an API key or account. The only exception is the optional Click Drop Finder report, which needs a Google Search Console service-account key if you choose to use it. The Query Explorer gives you a similar question list with copy and paste instead.
-
Where did the Georix screens go?
-
Since version 1.4 all Georix screens are in one Georix menu on the left of your admin area, starting with Start here. The screens themselves, their web addresses, and all your settings and data are unchanged.
-
Will this slow my site down?
-
No. The llms.txt content is generated and cached for 6 hours. Regular visitor page loads are completely unaffected — the plugin only runs when
/llms.txtor/llms-full.txtis specifically requested, not on every page load. -
Does it work with Elementor, Divi, or other page builders?
-
Yes. The file generation, schema output, AI Crawlers tab, and GEO Content Audit all work with any page builder. The FAQ and HowTo Gutenberg blocks require the WordPress block editor. The FAQ Question Suggester works on Elementor pages — it shows suggested questions in a copy box so you can paste them into your page builder’s FAQ widget.
-
Does it work with the Classic Editor?
-
The FAQ and HowTo blocks and FAQ Question Suggester require the block editor. All other features — file generation, schema output, AI Crawlers, LocalBusiness wizard — work with any editor.
-
How long until AI engines start citing my site?
-
Typically 4–12 weeks after your llms.txt is live and your key pages have FAQ blocks. Adding FAQ blocks to your homepage and top service pages has the biggest impact. AI crawlers recrawl sites regularly — once your content is indexed by them, citations follow.
-
Does this affect my regular Google SEO?
-
The llms.txt and llms-full.txt files are ignored by Google’s main search crawler, and Google has confirmed it does not use llms.txt. The LocalBusiness, Organization, and Article JSON-LD the plugin outputs is standard structured data Google still reads and can reward in search. FAQ and HowTo markup remains valid but no longer produces rich results in Google Search (FAQ results were removed on 7 May 2026; HowTo was deprecated in 2023) — the FAQ and HowTo blocks are worth using for clean, quotable content structure, not for rich snippets.
-
Is the FAQ Question Suggester free?
-
Yes, completely free. It works offline with no API call — all niche-specific question and answer templates are stored locally inside the plugin.
-
Does it work with WPML or Polylang?
-
Yes. When WPML or Polylang is active with two or more languages, Georix automatically groups
/llms.txtand/llms-full.txtby language. Each section includes that language’s homepage, key pages, posts, categories, and correct permalinks. Per-language taglines are used when available. You can toggle this under Settings Georix. -
Does it work on multisite?
-
Not yet. Multisite support is planned for a future release.
-
Can I control which AI bots can access my content?
-
Yes. The AI Crawlers tab lets you individually allow or block GPTBot (ChatGPT), ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, Bytespider, cohere-ai, and FacebookBot. Changes are written directly to your robots.txt.
Atsauksmes
Autori un izstrādātāji
“Georix – GEO & AI Citation” ir atvērtā pirmkoda programmatūra. Šo spraudni ir veidojuši šādi cilvēki.
Līdzdalībnieki
Tulkot “Georix – GEO & AI Citation” savā valodā.
Vai jūs interesē attīstība?
Pārlūkojiet kodu, apmeklējiet SVN krātuvi vai abonējiet attīstības žurnālu, ko izveidojis RSS.
Izmaiņu žurnāls
1.4.2
Makes the AI Crawler Cost report actionable, not just informational. All existing features and settings are unchanged.
- New: Block a bot right from the Crawler Cost table. Each of the 8 managed bots now shows an Allowed/Blocked status and a one-click Block/Allow button next to it — no more switching to the AI Crawlers tab to act on what this report shows you. It updates the exact same robots.txt setting either screen controls, instantly, with no page reload.
- If robots.txt rules are currently turned off, the report now tells you so before you click Block, with a link to turn them on.
- Reworded the note about blocking to be accurate now that a real block action exists on this screen: it explains what the Block button does (robots.txt, respected by well-behaved bots) versus true edge-blocking (a CDN like Cloudflare, which stops the request from reaching your server at all).
1.4.1
Small fixes and clearer wording. All existing features and settings are unchanged.
- Renamed: “Lost to AI” is now called Click Drop Finder — the old name didn’t say what the feature actually does. Its screen, URL and stored data are unchanged; only the label changed.
- Fixed: Query Explorer’s default match mode now finds a selected word anywhere in the query, not only at the very start, so a query like “doctors remain closed on Sunday why?” is now correctly matched by the “why” checkbox.
- Fixed: if the same term was typed as a keyword and also ticked as a word (for example keyword “near me” with the “near me” checkbox also selected), the regex wrongly required that term to appear twice in the query. One occurrence now satisfies both.
- Fixed: a security-scanner warning about unsanitized input on the Query Explorer screen (the value was already sanitized before use; the code now sanitizes it in the pattern scanners expect).
- Clearer, more natural wording on the Start here screen, in the guided tours and on the Click Drop Finder’s Search Console connection panel — no change to what any feature does.
1.4.0
Easier to find, easier to start. All existing features and settings are unchanged — these are additions only.
- New: one Georix menu. All screens now live under a single “Georix” menu instead of seven separate items under Settings. Every screen keeps its existing address, so bookmarks and saved settings keep working.
- New: Start here screen. A three-step checklist that ticks itself off, plus a short guide to every section with a real example. Opens once after first activation.
- New: guided tours. A tour for each screen and a full tour through all sections, built on WordPress’s own help pointers. Tours never start on their own.
- New: Query Explorer. Build a Google Search Console regex from plain words — copy and paste only, no connection.
- New: “while you wait” tips on reports that have no data yet.
- Improved: menu links for Local Business and AI Crawlers open the right tab directly.
- Fixed: the “Open Live File” buttons on the llms.txt preview tabs now open your live files (previously they had no link and could show a PHP warning).
- Readme rewritten around what Georix does for you, with the features that need no account listed first.
1.3.0
Win the answer. All existing features are unchanged — these are additions only. Removing the new feature files returns the plugin to its previous behaviour.
- New: Answer Capsule Engine — a separate editor panel (the existing GEO Content Audit and Answer Booster panels are untouched) with a target-question field and a Liftability Check that runs six deterministic checks against your rendered content: direct-answer presence (a 25–90 word paragraph in the first ~600 words carrying at least 60% of the question’s meaningful terms), self-containment, specificity, question-form heading, position, and freshness. The output is a red/amber/green verdict plus the exact reason each check failed — no composite score, because a made-up number would not mean anything.
- New: Answer Capsule block — publishes one question heading, one concise answer passage and up to three supporting facts as clean semantic HTML. Live 40–80 word counter and live self-containment linting warn but never block. The “start from what’s on the page” helper pre-fills the answer field with a paragraph you already wrote; the plugin never writes content for you.
- New: Settings Answer Capsules — a worklist of every published post and page showing its target question, last liftability verdict, whether it has a capsule, and when it was last modified, sortable and paginated.
- New: Lost-to-AI Question Finder (Settings Lost to AI) — opt-in, off by default. Connect your own Google Search Console property with a service-account key and see the question queries you appear for, a CTR-collapse detector, and a ranked action queue that deep-links each opportunity into the Answer Capsule editor with the question pre-filled. New table
georix_gsc_daily, daily sync cron, 12-hour cache, and a manual Refresh button. - Honesty: the CTR-collapse finding is labelled an inference, not a certainty, with its full method stated on screen — Google’s API does not report AI Overview presence. The “what changed since your last Answer Capsule” panel is labelled “what changed since”, never “results of”.
- Schema: the Answer Capsule block outputs no JSON-LD in this release. Emitting it would have meant changing the existing schema output, which this release deliberately does not touch. Noted for a future version. No rich-result claim is made anywhere.
- Privacy: the Search Console feature sends only a date range and requested dimensions to Google, for your own property, on a scheduled task or an explicit Refresh click — never on a visitor page load. Credentials are encrypted at rest where sodium or openssl exists, and the UI says plainly when neither does. Disconnect deletes them immediately.
- Uninstall additions are opt-in: cached verdicts (derived data) are always removed, your target questions and stored Search Console data are preserved unless you tick the removal box. Google credentials are always deleted.
1.2.0
The honest instrument panel for AI bot traffic. All existing features are unchanged — these are additions only. Removing the new feature files returns the plugin to its previous behaviour. Every number this release shows is derived from your own server data; estimates are labelled as estimates.
- New: AI Crawler Cost Report (Settings AI Crawler Cost) — measures AI crawler requests, their share of your total measured traffic, an estimated bandwidth figure (always labelled “estimated”), and the crawl-to-visit ratio joined to AI Traffic Insights. Includes Googlebot on purpose, a dependency-free trend chart, per-bot table, CSV export, and published industry benchmarks shown only as external context. New tables
georix_bot_dailyandgeorix_traffic_daily; capture is a single aggregate upsert per request. - New: Verified Bot Identity — confirms crawlers with forward-confirmed reverse DNS plus bundled IPv4/IPv6 ranges, on a scheduled task only (never during a page load). Raw IPs are used solely for the check and purged the moment it completes; only the verdict and resolving organisation are kept. Results use “unverified”, never “fake”.
- New: Reality Check (Settings Reality Check) — reports the truth about your own setup: whether anything actually requests your llms.txt, whether crawlers you disallowed are still crawling (with a Cloudflare edge-blocking recommendation and cache-lag awareness), and whether a page builder is hiding your content from crawlers behind JavaScript.
- Correction: full-page caching is now detected and both report screens warn that cache-served hits cannot be counted, so figures are an honest minimum.
- Correction: FAQ/HowTo claims repositioned as content structure, not rich results — Google removed FAQ rich results (7 May 2026) and deprecated HowTo (2023). Blocks and schema output are unchanged; only the wording changed.
- Correction: the Crawler Analytics tab is renamed Discovery File Requests to reflect that it shows only /llms.txt-family requests, not sitewide crawling.
- Correction: llms.txt is presented plainly — Google has confirmed it does not use it; we generate it because it is free and may matter later, and Reality Check tells you if anything requests it.
- Correction: uninstall now cleans up all v1.1/v1.2 options and, when you opt in, the new tables. Data is preserved by default.
- We measure; we do not block. For blocking, we recommend Cloudflare, which does it at the edge for free. No new external services, no API keys, nothing phones home.
1.1.0
This is a proof-and-action release: Georix now shows you real results and helps you improve, instead of only generating files. All existing features are unchanged — these are additions only. Removing the two new feature files returns the plugin to its previous behavior.
- New: AI Traffic Insights — first-party analytics showing which pages get real click-throughs from AI engines (ChatGPT, Perplexity, Gemini, Claude, Copilot, and more), read from your own site’s traffic. Scoreboard, trend chart, top-cited-pages table, opportunities panel, CSV export, and configurable retention.
- Privacy-first storage: AI source, landing path, UTC day, and count only — no IP addresses, no raw user agents, referrer query strings stripped. Nothing is transmitted off your server. Data is preserved on uninstall unless you opt in to removal.
- Honest measurement: the UI names its blind spots (Google AI Overviews and no-click answers) and presents the figure as a verified minimum, not a total.
- New: Answer Booster — a separate editor panel (the existing GEO Content Audit sidebar is untouched) that flags unsourced statistics, undated time-sensitive claims, filler section openings, and missing quotable claims, with safe click-only template fixes that never rewrite your words or invent a source. Per-check and master toggles. 100% offline.
- Reposition: llms.txt is now framed as agent-readiness infrastructure, with the evidence-backed AI Traffic Insights and Answer Booster as the core.
- No new external services, no API keys, no data leaves your site.
1.0.1
- Fixed fatal error on activation when Rank Math, SEOPress, or AIOSEO stores robots meta as an array instead of a string
1.0.0
- Initial public release
- Auto-generates /llms.txt and /llms-full.txt served live via WordPress URL rewriting — no file uploads needed
/ai.txtAI usage permissions file with admin toggle and policy settings/llm.txtcompatibility alias (same content as llms.txt)<link rel="alternate">discoverability tags in site<head>- LocalBusiness Schema Wizard with 100+ schema.org types including India-specific business categories
- AI Crawler Manager — per-bot Allow/Disallow for 8 AI crawlers written to robots.txt, with 40+ bot registry, robots.txt conflict detection, and expanded directives
- Custom Sections builder — add your own sections and links to llms.txt from the admin UI
- FAQ Gutenberg block with FAQPage JSON-LD schema
- HowTo Gutenberg block with HowTo JSON-LD schema
- FAQ Question Suggester with niche detection — dental, legal, medical, accounting, restaurant, generic
- GEO Content Audit sidebar — live 9-check score updating in real time in the block editor, with Elementor-aware word count and heading detection
- Organization, Article, BreadcrumbList, and speakable schema output site-wide
- EEAT and AEO Readiness Score — 5-pillar 0–100 score
- WPML and Polylang support — llms.txt and llms-full.txt grouped by language with per-language URLs, taglines, key pages, and content
- Crawler visit logging (database table, daily cleanup cron, Crawler Analytics tab)
- llms.txt Health Panel with HTTP status, cache, permalink, and conflict checks
- Physical file conflict detection and admin warnings
- SEO plugin noindex filtering (Yoast, Rank Math, SEOPress, AIOSEO)
- Per-post include/exclude controls and content selection modes (translation-aware)
- Intelligent content filtering with configurable minimum word count
- Quality-scored post ordering for better llms.txt output
- Check File Status tests all active discovery URLs
- Documentation tab with step-by-step guidance built into the plugin
- All features fully unlocked — no license keys, no Pro tier, no usage limits