You can't buy a citation.
You can stop being skipped.
Answer engines increasingly end the search before anyone clicks. Citely measures the four things that decide whether your content can be the source an answer is built from — crawler access, extractability, structured data, and a machine-readable summary — and reports what it finds as counts you can check.
- Reports counts, not promises
- Audit, llms.txt & crawler controls free forever
- Grounded in published GEO research
- capabilities built and tested
- 9capabilities built and tested
- specified, not built yet
- 4specified, not built yet
- AI crawler access control, per agent, with training and retrieval separated
- Extractability audit across six signals, scored 0–100 per page
- Site-wide visibility report: median score, per-weakness counts, weakest pages
- AI crawlers controlled per agent
- 6AI crawlers controlled per agent
- Extractability score per page
- 0–100Extractability score per page
- Audit, llms.txt & crawler controls
- FreeAudit, llms.txt & crawler controls
- Free and open source
- GPLFree and open source
Why Citely
The honest version of
answer-engine optimisation.
This is a field full of unfalsifiable claims. The way out is to only ever talk about what can be counted from your own site — and to say plainly where the measurement stops.
Training and retrieval are not the same thing
Blocking GPTBot is an editorial choice about model training. Blocking OAI-SearchBot removes you from the live answers you were trying to appear in. Most sites get this backwards, so Citely records a purpose for every agent and flags blocked retrieval crawlers as a misconfiguration rather than a preference.
Every claim is a number you can check
How many pages have no liftable answer block. How many cite no external source and quote no statistic. Which crawlers you block. Whether you have an llms.txt. Those are countable from your own database, which is why they are what this product talks about.
Grounded in the published research
The signals are weighted toward what Aggarwal et al. found in GEO: Generative Engine Optimization (KDD 2024) — citing sources, adding quotations and adding statistics produced the largest visibility gains among the interventions they tested. That is a real finding on a specific benchmark, not a guarantee about your site.
Honest about the ceiling
Nothing here can tell you whether ChatGPT cites you, because that is not measurable from inside WordPress. What it can tell you is whether you have given an engine a reason not to — and that is the part you actually control.
What is built today
Everything that measures, and everything free
Each capability below is implemented and tested in the plugin as it stands. The metered half is listed further down, marked as unbuilt.
Per-agent AI crawler control
Six crawlers across four vendors — OpenAI, Anthropic, Perplexity and Google — each recorded with whether blocking it affects training or live retrieval. Directives are emitted as denials only, so a per-agent rule never narrows a wildcard defined elsewhere in your robots.txt.
Extractability audit, per page
Scores each page 0–100 across six signals: whether it opens with a liftable answer, whether it cites external sources, whether it quotes statistics, whether headings ask questions, whether it exposes a modified date, and whether it is too thin to be worth quoting.
Site-wide visibility report
Median score, a count per weakness, and the weakest pages ranked — so the next thing to fix is the page a fix will move, not whichever one you happened to open.
llms.txt, generated from real content
Both llms.txt and llms-full.txt, built from what the site actually publishes rather than hand-maintained. Worth knowing: llms.txt is a proposed convention and no major answer engine has confirmed it as a retrieval or ranking input. It is cheap insurance, not a lever, and this plugin will not tell you otherwise.
Score decomposition, not a number to trust
The weight of every signal is exposed, so the admin screen can break a score down instead of asking you to take it on faith. A score a reader cannot decompose is one they have to believe, and this product does not get to publish those.
Admin screen and REST surface
The visibility report has a screen in wp-admin and authenticated REST routes behind real capability checks, so the audit is scriptable as well as readable.
How it works
Three steps that work, one that doesn't yet
Shown end to end so the shape is clear — with the unbuilt step marked rather than written in the present tense.
- 01
Check who you're blocking
Citely reads your crawler access per agent and separates training from retrieval. If you are blocking a retrieval crawler, that alone can keep you out of live answers while everything else is right.
- 02
Audit what you publish
It scores each page for extractability — liftable answer, citations, statistics, question headings, freshness, substance — and rolls that into a site-wide report with the weakest pages ranked.
- 03
Publish a machine-readable summary
llms.txt and llms-full.txt are generated from your real content. A proposed convention rather than a proven ranking input, so treat it as cheap insurance.
- 04
Rewrite what the audit flagged
Turn a flagged opening paragraph into a liftable answer block. Every rewrite arrives as a word-level diff in a review queue — never applied silently — splices back without breaking block markup, and can be reverted from the journal.
Try it without installing anything
Free tools and guides
The checker runs the same analysis as the plugin's audit, against a public URL instead of your posts table. No signup.
Build status
What ships, and what doesn't
Including the things this product will not do at all, which in a category this noisy is the more useful list.
Citely is not released — there is no download yet — but both halves now exist. Everything free is implemented and tested: crawler controls, the extractability audit, the site-wide report, llms.txt and per-page structured data. So is the metered half: answer-block rewriting, with a word-level diff, a review queue and a revert. What is still unbuilt is listed beside it, because that list is the reason to trust the rest.
Implemented and tested
- AI crawler access control, per agent, with training and retrieval separated
- Extractability audit across six signals, scored 0–100 per page
- Site-wide visibility report: median score, per-weakness counts, weakest pages
- llms.txt and llms-full.txt generated from real content
- Per-page structured data, where no SEO plugin already provides it
- Answer-block rewriting, reviewed as a word-level diff before anything is applied
- Applying a rewrite without breaking block markup, and reverting a run
- Admin screens and REST routes
- Licensing, metered in items, on the shared WP Shelf platform
Specified, not built
- Heading rewriting toward question form — the service accepts the item type, the plugin does not offer it yet
- Generated FAQ schema
- Citation tracking — whether you are actually quoted by a given engine
- A public release: no WordPress.org listing and no download
Why there are no prices here
The audit, llms.txt and the crawler controls are free on every tier including with no licence — they are the reason to install, and metering them would gate the thing that makes the rest worth doing. Only rewriting consumes quota, metered in items, where one item is one rewritten answer block. There are no prices on this page because the ladder is not calibrated against measured per-item cost yet, and a quota published before it is measured is a guess with a number on it.
FAQ
Questions, answered
Citely is a WordPress plugin for AI visibility: making a site legible, accessible and quotable to answer engines. It does four things — controls AI crawler access per agent, audits whether pages are extractable, generates a machine-readable llms.txt summary, and reports structured-data completeness. As of plugin version 0.24.0 all of that is implemented and tested, along with answer-block rewriting and revert. Heading rewriting and citation tracking are not built, and there is no public release yet.
No plugin can promise that, and Citely does not. Whether a given engine quotes a given page is not measurable from inside WordPress, so any product claiming it is asking you to take an unfalsifiable claim on trust. What Citely does is subtractive: it finds the reasons an engine would skip you — a blocked retrieval crawler, a page with no liftable answer, no external citations, no statistics, no machine-readable summary — and reports each as a count you can verify yourself.
Not yet. Citely is not released and there is no download or WordPress.org listing. The plugin is further along than that implies — crawler controls, the audit, the site-wide report, llms.txt and answer-block rewriting are all implemented and tested — but until it ships there is nothing to install. The AI Visibility Checker on this site runs the same analysis against any public URL in the meantime.
GPTBot collects content for model training; blocking it excludes your content from training data. OAI-SearchBot fetches pages for live retrieval; blocking it removes your site from ChatGPT's live search results. They are different decisions with opposite consequences, and conflating them is the single most common way a site becomes invisible to answer engines while doing everything else right. Citely records a purpose for each of the six crawlers it manages and flags a blocked retrieval crawler as a misconfiguration.
Not provably, yet. llms.txt is a proposed convention published at llmstxt.org, and no major answer engine has confirmed it as a retrieval or ranking input. Citely generates it because it costs almost nothing and, if adoption arrives, a curated index already exists instead of whatever a crawler would otherwise infer. Treat it as cheap insurance rather than a lever — and be sceptical of anything that sells it as more than that.
Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024), which tested a range of content interventions and found that citing sources, adding quotations and adding statistics produced the largest visibility gains. Citely's audit weights those signals accordingly. It is a real result on a specific benchmark, not a guarantee that the same interventions will move your traffic.
The audit, llms.txt and the crawler controls are free on every tier, including with no licence — they are the reason to install it. Only rewriting is metered, in items, where one item is one rewritten answer block. The prices are not published because the item counts have not been calibrated against measured per-item cost, and publishing a quota before measuring it is guesswork with a number attached.
No. It does no sitemaps, no redirects, no rank tracking and no keyword research, and it does not touch titles, meta descriptions or alt text — those belong to Metaly, a separate WP Shelf plugin. Citely is about whether a machine reading your page can extract an answer from it and is allowed to fetch it at all.
WordPress 6.4 or later and PHP 8.1 or later, the same baseline as every WP Shelf plugin. Citely is GPLv2-licensed.
Check a page right now.
The plugin is not released yet, but the analysis behind it already runs in your browser against any public URL. That is the honest way to find out whether any of this applies to you.