Coywolf SEO is built on a simple idea: an SEO plugin should give you exactly what you need, and nothing more. No upsells, no nags, no bloat, no features you have to turn off \u2014 just the essentials, done right, with the plugin doing the heavy lifting in the background.<\/p>\n\n
Features:<\/p>\n\n
Privacy-first: this plugin includes no analytics, no tracking, and no data gathering \u2014 nothing about you, your site, or your visitors is ever collected. Outbound connections happen only for features you turn on, and each is detailed under \"External services\" below: with IndexNow enabled, the changed URL is sent to Microsoft Bing's IndexNow endpoint on publish, update, and delete; with AI enrichment or Image Text enabled, the post title and content \u2014 and, for Image Text, the image itself \u2014 are sent to the AI service you choose (Anthropic, OpenAI, or Google Gemini) using your own API key, and extracted entity names are looked up on Wikidata's public API; and with the Link Manager enabled, the plugin requests the URLs you have linked to in your own content to check whether they still work. Nothing else, nowhere else.<\/p>\n\n
This plugin can connect to the third-party services below. Each is optional and is contacted only when you enable the feature that uses it; none are contacted on a default install.<\/p>\n\n
AI provider \u2014 Anthropic, OpenAI, or Google Gemini (you choose one)<\/p>\n\n
Used by AI Schema enrichment, AI meta descriptions, and Image Text. When you enable an AI feature and supply your own API key, the plugin sends content to the single service you selected in Settings:<\/p>\n\n
Requests are authenticated with the API key you provide and are sent only to the one service you select. Terms and privacy policy for each:<\/p>\n\n
Wikidata (Wikimedia Foundation)<\/p>\n\n
Used by AI Schema enrichment to ground entities to real identifiers. The entity-name strings extracted from your content are sent as lookups to Wikidata's public API (https:\/\/www.wikidata.org\/w\/api.php); no API key and no personal data are involved. Terms: https:\/\/foundation.wikimedia.org\/wiki\/Policy:Terms_of_Use \u2014 Privacy: https:\/\/foundation.wikimedia.org\/wiki\/Policy:Privacy_policy<\/p>\n\n
IndexNow (Microsoft Bing)<\/p>\n\n
Used by the IndexNow feature. When enabled, the URL of a post or page you publish, update, or delete \u2014 together with the auto-generated site key \u2014 is submitted to Bing's IndexNow endpoint (https:\/\/www.bing.com\/indexnow) so search engines can recrawl it promptly. No personal data is sent. About IndexNow: https:\/\/www.indexnow.org\/ \u2014 Microsoft Privacy Statement: https:\/\/privacy.microsoft.com\/privacystatement<\/p>\n\n
Link checking (the sites you link to)<\/p>\n\n
Not a third-party service, noted here for transparency: when the Link Manager is enabled, it sends HTTP HEAD (and, if needed, GET) requests to the URLs you have linked to in your posts and pages to record each link's HTTP status. Those requests go to the sites you chose to link to \u2014 never to Coywolf or any other service \u2014 and carry only an ordinary request with a plugin User-Agent; no information about your site or your visitors is transmitted.<\/p>\n\n\n
Titles are composed from your settings, with the site name appended only where you enable it (a force-rewrite option handles themes that build their own title tag). Meta descriptions are never auto-generated \u2014 the homepage uses the Tagline, posts and pages use their manual Excerpt, terms use their description \u2014 unless you turn on AI meta descriptions or exclude meta descriptions entirely.<\/p><\/dd>\n
Each page outputs one JSON-LD graph, but pages default to no Article type. Set a default in Site Details \u2192 Pages, or override it per page in the SEO panel.<\/p><\/dd>\n
Bring your own API key for Claude (Anthropic), OpenAI, or Google Gemini and pick one in Settings (or define it in wp-config.php). The AI features require WordPress 7.0+ because they run on its bundled AI client.<\/p><\/dd>\n
No. The model only extracts entity names; the real items are looked up on Wikidata's public API, the model chooses among them, and the chosen item's type is verified \u2014 so identifiers are never fabricated.<\/p><\/dd>\n
It runs through your service's Batch API at roughly half the standard token price, in the background, and can take up to an hour. Posts already analyzed with the current settings are skipped, so re-running is inexpensive.<\/p><\/dd>\n
No \u2014 detected entities and generated descriptions and image text are kept and still used. Only your saved API key is deleted (re-enter it when you turn AI back on; a key in wp-config.php is ignored while off).<\/p><\/dd>\n
No. It processes in the background on WP-Cron, so you can leave and come back to check progress. On a very low-traffic site, leaving the tab open helps it along.<\/p><\/dd>\n
Bulk image text defaults to the cheaper Batch API (about half price; results within the hour, up to 24 hours). Tick \"real-time\" for immediate processing at the standard rate. Gemini has no vision Batch API, so it always runs in real time.<\/p><\/dd>\n
Alt text is intentionally left empty for purely decorative images (an accessibility best practice). Title, caption, and description are always written.<\/p><\/dd>\n
Two things, each only when an uploads-folder image exactly matches a Media Library item: it adds the missing attachment ID to in-content images, and it converts Custom HTML or classic image figures into real image blocks. Run Preview first \u2014 it edits post content \u2014 and note that converting drops any custom inline figure styling.<\/p><\/dd>\n
Propagation only updates core\/image blocks that reference the file, and only empty fields unless \"Overwrite\" is on. Images added as Custom HTML or by URL aren't matched until you run \"Fix missing image IDs\" first.<\/p><\/dd>\n
The moment you delete a published post or page, the Redirects screen asks what to do with its URL \u2014 mark it gone (410), redirect it, or dismiss. Pending decisions also wait on the Redirects page.<\/p><\/dd>\n
Yes \u2014 from the Redirection plugin (even when it's deactivated, read straight from the database) and from Yoast SEO Premium. Duplicates are skipped, so importing is safe to re-run.<\/p><\/dd>\n
Only the first analysis is a full scan; after that the inventory keeps itself current as you create, edit, and delete posts. A Throughput setting tunes how hard it scans.<\/p><\/dd>\n
It runs on a PHP port of Google's open-source Robots.txt Parser and Matcher Library \u2014 the Robots Exclusion Protocol (RFC 9309) reference implementation Googlebot uses \u2014 so the Test URL result, the conflict warnings, and the redundancy checks all reflect exactly how Google will interpret a rule, including * wildcards, $ end-anchors, percent-encoded paths, and Allow-vs-Disallow longest-match precedence. What the plugin previews is what Googlebot will do.<\/p><\/dd>\n
Virtual (the default) means WordPress serves robots.txt with your managed rules injected \u2014 no file is written. Physical mode writes a real file and preserves your hand-written lines. On activation, the manager imports and tidies any existing robots.txt.<\/p><\/dd>\n
Yes. Turning the Robots.txt Manager off \u2014 or deactivating the plugin \u2014 prompts you to restore your original robots.txt or keep the managed rules.<\/p><\/dd>\n
It's built when the page is served, not when you save, so headings added later (including by shortcodes and synced patterns) are always included. Each heading has an \"Exclude from table of contents\" toggle, and a minimum-headings threshold keeps the table off short posts.<\/p><\/dd>\n
It's true art direction: your phone-specific image is served below 768px through a media query, so it's guaranteed on small screens rather than left to the browser's srcset guesswork. The breakpoint is filterable.<\/p><\/dd>\n
No. While the other plugin is active, Coywolf SEO suppresses its front-end output and edit-screen boxes through that plugin's own switches, so nothing is duplicated as you migrate \u2014 but it leaves their sitemaps and redirects running.<\/p><\/dd>\n