Publish an llms.txt Map So AI Models Can Find What to Cite on WordPress

robots.txt tells crawlers what they are allowed to fetch. It assumes the bot already knows your site exists and which URLs matter. An answer engine often does not. It needs a short, readable map of the pages you actually want cited: what they cover, why they matter, where to start.

That map is llms.txt. Plain text. Written for models, not browsers. Rankify publishes it as a readable map of your site, serves it virtually with robots.txt (nothing left behind on the server), and pairs it with labelled AI crawler control so access policy and the map stay in one place.

This post is about shipping that map on WordPress. Crawler labels get a light mention because a map without fetch access is theatre. Titles, schema graphs, sitemaps, answer blocks, and referral dashboards live elsewhere in Rankify. They are not the hero here.

For the wider “are assistants using us?” funnel, see how to tell if AI answer engines are using your WordPress pages. For the training vs citation policy line, see block AI training crawlers without erasing ChatGPT citations.

The problem: crawlers get rules, models need a map

Classic SEO taught us to polish robots.txt, XML sitemaps, and meta robots. That still matters. Googlebot and friends already know the grammar. Language models browsing for citation candidates often look for something simpler: a human-readable index of important URLs with short summaries.

Without that file, a model that is allowed to fetch you still has to guess which of your 400 posts is the durable guide and which is a 2019 changelog. Sitemaps list URLs. They do not explain intent. Your homepage nav explains intent to people. Models do not click nav the way a shopper does.

So you end up with a site that is “open” in robots.txt and still hard to cite, because nothing pointed the model at the money pages, the comparison posts, or the docs you actually want named in an answer.

What llms.txt is (and is not)

llms.txt is a conventional plain-text file at the site root. Think of it as a curated table of contents for language models: key URLs, short descriptions, maybe groupings by topic. It is not a secret ranking API. It is not a substitute for allowing citation crawlers. It is not a second sitemap with fancy XML.

It is a readable map. Models that look for it can follow it. Models that ignore it cost you almost nothing to serve. Rankify’s product line is blunt about that: llms.txt published: a readable map of your site written for models.

Rankify also serves robots.txt and llms.txt virtually, so nothing is left behind as a messy static file fight with your host, CDN, or another plugin. The plugin owns the response. You do not babysit a file on disk that drifts after the next deploy.

What Rankify actually ships for the map

Stick to real product surface. No invented “AI rank score from llms.txt” claims.

llms.txt published. A readable map written for models, kept current as you choose what belongs in it.

robots.txt and llms.txt served virtually. No leftover files on the server. Rankify answers those paths itself.

AI crawler control on free. Sixteen crawlers across ten vendors, each explained in the UI. Training vs citation separated and labelled, so you can allow the bots that cite without treating every bot string as the same threat.

Free includes the map layer. Free Rankify already covers llms.txt, robots.txt, and AI crawler control (plus the rest of the free SEO baseline: titles, sitemaps, schema, redirects, content analysis, and a short AI crawler log). You do not need Pro just to publish the map.

Content analysis runs locally. Keyword, structure, and readability checks stay in the browser. No upload, no API key for that path. Useful when you tighten the pages the map points at.

Titles, schema, and sitemaps still ship in Rankify. Mention them only as the classic half. Today’s job is the map models can follow.

How to publish a useful llms.txt on WordPress

Treat the map like a product homepage for machines, not a dump of every URL.

1. Turn the map on in Rankify. Open the AI and GEO settings. Confirm llms.txt is published. Leave it on by default unless you have a deliberate reason to hide the site from that convention.

2. Curate, do not dump. Include cornerstone guides, pricing or product pages you want cited, comparison posts that state facts cleanly, and docs that answer real questions. Exclude noindex, thin tags, login, cart, and “hello world” leftovers the same way you would from a sitemap.

3. Write short summaries that stand alone. One or two sentences per URL. Name the subject. Skip “click here” and pronoun-heavy blurbs that only make sense after three prior paragraphs.

4. Align the map with access policy. If citation crawlers are blocked site-wide, the map will not save you. Use Rankify’s labelled list: sixteen crawlers, ten vendors, training vs citation called out before you flip Allowed or Blocked. Light rule of thumb: keep citation paths open if you want to be named; block trainers only when that is deliberate policy. Deeper walkthrough: training vs citation without collapsing both into one deny.

5. Verify the live path. Open https://yoursite.com/llms.txt in a browser. Confirm the map loads, summaries look sane, and Rankify is the one serving it (virtual serve, not a conflicting static file).

6. Point the map at pages worth quoting. After access and the map are solid, tighten the destinations: answer early, question-shaped headings, specific claims. Content analysis in the editor runs locally if you use that check. Measuring who came back from answers is a separate post: measure WordPress visits from ChatGPT and Perplexity.

Pitfalls that waste the map

1. Publishing an empty or auto-everything map. If every tag archive and author page is listed with a one-word blurb, models get noise. Curate.

2. Hand-editing a static llms.txt next to Rankify. Two owners fight. Prefer Rankify’s virtual serve so the live file matches the settings UI.

3. Blocking every AI user-agent, then wondering why the map “does nothing.” Access first, map second. Labels exist so you do not erase citation paths while refusing training.

4. Treating llms.txt as a ranking hack. It is a discovery aid for models that look for it. It does not replace titles, crawlability, or clear writing.

5. Letting the map rot. When you ship a new comparison post or retire a guide, update what the map promotes. Stale maps point models at the wrong story.

6. Promising “ChatGPT will cite us because we have llms.txt.” No honest plugin can guarantee a citation. Rankify publishes the map and controls labelled crawlers. The quote still depends on whether your page is the better source.

How this fits DaftPlug Pro

Rankify is the SEO and GEO layer in the suite. Free already publishes llms.txt, serves robots.txt virtually, and ships labelled AI crawler control. Pro adds deeper answer-engine measurement and structure tools when you need them. You do not buy a second SEO product just to get the map.

One Pro licence unlocks Progressify (PWA), Lightify (performance), Generatify (AI), Rankify (SEO and GEO), and Snapshotify (backups). Plans differ by site count only: Pro $37.99/year (1 site), Pro Plus $74.99 (5), Pro Max $138.99 (100). Features are identical across plans. Checkout is Freemius, with a 14-day money-back guarantee. See pricing.

Takeaway

robots.txt is the rulebook for crawlers that already know the grammar. llms.txt is the readable map that tells models which pages you want cited and why they exist. Publish it on WordPress with Rankify, keep robots.txt and llms.txt virtual so nothing rots on disk, and pair the map lightly with labelled training vs citation crawler control so fetch access matches intent. Curate the URLs, verify /llms.txt live, then improve the pages the map points at. That is how you stop hoping models guess your best work and start handing them a map they can follow.