Mephivio Robots LLMs

The Exclusions tab offers an Invalider action on each row to lift an ordinary exclusion for an eligible published item. Protected items and WooCommerce system pages cannot be overridden. Regeneration is requested after invalidation.

Mephivio Robots LLMs generates and publishes machine-readable discovery files that can help compatible crawlers understand your public content. llms.txt is an emerging convention and is not guaranteed to be consumed by every AI or search crawler:

• llms.txt — a Markdown index of your public content, in the format expected by the llmstxt.org specification.
• llms-full.txt — the full text corpus of your content, cleaned up (removal of technical noise from page builders, scripts, styles, shortcodes…).
• llms-schema.json — the same index in JSON-LD format (schema.org ItemList), for agents able to consume structured data.
• robots.txt — a granular policy per bot category (search engines, AI, social networks, “bad bots”), with your custom rules and sitemaps.

What makes this plugin different

Mephivio Robots LLMs is designed as a local, auditable discovery-control layer rather than only an llms.txt writer. In one interface it combines multilingual document generation, WooCommerce catalog indexing, robots.txt policy controls by crawler family, per-content exclusions and visibility scoring, diagnostics, previews, and privacy-friendly bot request statistics. It does not require an external API or send site content to a third party.

Key features

• Admin interface language switch: English or French, with English as the default, independently of the WordPress locale.
• Smart content extraction: Gutenberg, Elementor, Divi, WPBakery, Beaver Builder, FAQ/accordions, WooCommerce products.
• LLM visibility score per content item (0 to 100), with a configurable minimum threshold.
• Per-post/page exclusion metabox, respects noindex (Yoast SEO, Rank Math, All in One SEO).
• Live preview of generated files directly in the admin.
• URL inclusion tester and generation logs.
• Dynamic (virtual, no disk writes), physical, or hybrid publishing.
• Delta mode: only reprocesses content that changed since the last generation.
• Polylang / WPML compatibility: one document per detected language (llms-fr.txt, llms-en.txt…) in addition to the default document.
• Aggregated, privacy-friendly statistics of known bot requests on your llms.txt / llms-full.txt / robots.txt files.
• CSV export of excluded content, with the exclusion reason.
• WooCommerce-compatible (products are selected by default when WooCommerce is active, but can be disabled in the content-type settings; cart/checkout/account pages are excluded).
• Per-bot AI crawler control, grouped by purpose (model training, AI search and answers, user-triggered fetch): for example block GPTBot while keeping OAI-SearchBot and ChatGPT-User. The robots.txt output follows RFC 9309 (global rules repeated in every group).
• Markdown version of each indexed page (/my-page.md), as suggested by the llms.txt proposal (opt-in).
• llms.txt “## Optional” section (by content type or below a score) and optional taxonomy archive sections (categories, product categories…).
• “LLM visibility” panel in the block editor: exclusion switch, custom LLM summary used in llms.txt and JSON-LD, current score.
• WooCommerce price and availability in llms-full.txt, the JSON-LD feed (schema.org Offer) and Markdown pages.
• Daily charts of bot requests and an aggregated count of human visits coming from AI assistants (ChatGPT, Perplexity, Claude, Gemini, Copilot…).
• Configuration check: permalinks, llms.txt format, documents blocked by robots.txt, physical files hiding the dynamic ones, sitemaps, WP-Cron, robots.txt conflicts with SEO plugins.
• Automatic sitemap detection (WordPress core, Yoast SEO, Rank Math, All in One SEO, SEOPress) and backup of any existing robots.txt / llms.txt before it is replaced in physical mode.

Privacy

The plugin does not send any data to a third-party service. Bot traffic statistics are stored locally only (aggregated by bot and by day, without storing raw IP addresses or user-agent strings) and can be disabled or reset at any time from the settings.

Visits coming from AI assistants are counted the same way: only a counter per day and per assistant, and a counter per landing path, detected from the Referer header or the utm_source parameter. No IP address, user-agent or cookie is stored; data older than 30 days is deleted. This tracking can be disabled in Settings > Performance. Pages served by a full-page cache do not run PHP, so part of these visits is not counted.

UTF-8 publishing (1.6.3)

UTF-8 publishing (1.6.3)

Dynamic llms.txt and llms-full.txt responses use text/plain; charset=utf-8. Physical publication uses the WordPress Filesystem API without extra encoding conversion. Invalid UTF-8 output aborts publication while preserving previous files. After upgrading, use Regenerate now and clear any page/CDN caches. Already-corrupted source content needs to be corrected at its source. The supported filename is llms-full.txt (not llmsfull.txt).