| 开发者 | havoric |
|---|---|
| 更新时间 | 2026年11月15日 21:01 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
llms.txt is a plain-text index of your content at a fixed address, so an AI engine can find what matters without crawling everything. Citolume builds yours from your published content and serves it at /llms.txt: nothing written to disk, updated whenever you publish. Add site context, write your own sections (they survive regeneration), and give each post its own description or leave it out.
Honest about the limits
llms.txt is a proposal, not a standard. Not every AI engine reads it. It costs nothing to publish, but treat it as housekeeping, not a guarantee./wp-content/plugins/citolume.llms.txt needs pretty permalinks; on plain permalinks the plugin says so and links to the setting.That is exactly the question Citolume answers. Add the questions your customers would ask, and each scan asks the AI engines on your plan — Gemini, Perplexity and ChatGPT — and records whether your site is among the sources each one cites, how often, and who is cited instead.
SEO plugins optimise your pages for Google's search results. Citolume measures something they do not: whether AI answer engines cite you. It has no meta tags, sitemaps or schema of its own, so it never conflicts with your SEO plugin — use both.
Yes, to start. The crawler log and llms.txt are free with no connection at all. Connecting is free too — one click, no card — and adds AI readiness and citation tracking for two questions, scanned monthly on Gemini. Paid plans add questions, engines and weekly scans; billing lives in your account at app.citolume.com, never inside WordPress.
Because they are, if you run page caching. Caching plugins serve a saved copy of a page without starting WordPress, so the plugin never sees those requests. Anything cached is invisible to it, and the counts should be read as a floor rather than a total.
There is no file. It is generated when requested and served at /llms.txt. That means it never goes stale, and it works on hosts where the web root is not writable. If you already have a real llms.txt file in your site root, that one is served instead and the plugin leaves it alone.
Your site is using plain permalinks, so WordPress never receives the request. Pick any other option under Settings → Permalinks.
Citation tracking asks each engine through its official API, with web search on. That is a close proxy for the app you use, but not identical — the consumer apps have their own retrieval on top. When an engine answers a question without searching at all, that is reported honestly rather than counted as though you were not cited. Answers vary, so paid plans ask each question twice per scan (free asks once) and results are reported as a frequency, never a yes/no.
The crawler log, the crawler-block check and llms.txt work entirely inside your own site, with no account and no external service. Citation tracking talks to our service once you connect — an explicit click in the setup wizard, never automatic. From then on the plugin sends the site's licence key, the questions you chose to track, and once a day the crawler log's daily totals (crawler name, page path, date, count). If the service is unreachable, or you never connect, everything else carries on exactly as before.
No. One small write per AI crawler visit, and nothing at all for human visitors. Visits are aggregated per crawler, per page, per day rather than stored one row per request, so the table stays small. Records older than 90 days are removed automatically.
Not with this plugin. Its job is to show you what is happening. Blocking is done through robots.txt.
/llms.txt with its own screen: site context, custom sections that survive regeneration, per-post descriptions and an exclude switch.