Tool reference

Find `find` text across content and replace with `replace`. Set post_type='product' to run it on WooCommerce PRODUCTS (a product is a post) or any CPT; default 'post' = blog posts. dry_run=True (default) only REPORTS which items would change (does NOT edit). Run with dry_run=False to actually apply. Always dry-run first, show the user, then apply.

Schedule MANY posts at once. `schedule_json` is a JSON array of {"post_id": N, "datetime_iso": "2026-07-15T09:00:00"} (site-local time). Sets each to status=future. Returns per-post result.

Create an item in ANY custom post type (e.g. post_type='service'). content is HTML. `terms` assigns custom-taxonomy terms as 'taxonomy:id[,id];taxonomy2:id' - e.g. 'service_hub:12' puts it in the 'lessons' hub (find IDs with list_taxonomies). Use list_post_types to get the post_type slug first.

Create a new page. status defaults to draft. content is HTML (Gutenberg block markup works too).

Create a post - content is HTML, category_ids and tag_ids are COMMA-SEPARATED ID STRINGS like '3,7' (never arrays or ints), status defaults to draft, and you should always pass a short hyphenated `slug`. featured_media_id = media ID for the thumbnail. slug = a short, keyword-rich, hyphenated URL slug (e.g. 'watercolor-lighthouse-drawing', not the long auto-slug WordPress builds from the title).

Delete an item of ANY custom post type. Default = Trash (recoverable). permanent=True deletes forever.

Delete a PAGE. Default = move to Trash (recoverable). permanent=True = delete forever. Use list_pages to find the page ID. (For posts, use delete_post.)

Delete a post. Default = move to Trash (recoverable). permanent=True = delete forever.

Clone a post (great for templates): copies title, content, excerpt, categories and tags into a NEW draft. Pass new_title to rename the copy. Returns the new post id. For a CUSTOM post type (e.g. 'service'), pass post_type - the copy is created in the same type.

Find published posts with an EMPTY excerpt. With apply=False reports them. With apply=True, generates a ~30-word excerpt from the first paragraph of each (your theme uses excerpt for meta description).

Get one page's full raw HTML content + title, status.

Get one post's FULL raw content (HTML) + title, excerpt, status, categories, tags, featured image id. Read this before editing. For a CUSTOM post type (e.g. 'service'), pass post_type so it reads from the right endpoint - otherwise only normal posts work.

List items of ANY custom post type (e.g. post_type='service'). Returns id, title, status, link, slug + any taxonomy term ids. Use list_post_types first to find the post_type slug.

List pages (About, calculators, etc.). Returns id, title, status, link.

List every REST-enabled post type on the site (built-in AND custom, e.g. 'service', 'product', 'portfolio') with its slug, label, and REST base. Use this FIRST to discover custom post types before creating/editing their items with the *_cpt_item tools.

List posts. Filter by `search` text and `status` (publish,draft,pending,private,future). Returns id, title, status, link, categories, tags. Use to find a post ID before editing.

List a post's saved revisions (id, date, snippet). Use restore_revision to roll back if an edit went wrong.

One-shot publisher: creates a post AND (optionally) sets author, categories, tags, excerpt, a short keyword-rich `slug`, generates+attaches a Gemini featured image, and returns suggested internal links for the topic so you can weave them in. `content_html` should be the full article HTML (include your schema script if you have one). ALWAYS pass a short `slug` (3-6 words) for a clean SEO URL. Use status='draft' to review first, 'publish' to go live.

Fetch a PUBLIC web page (e.g. a reference article the user linked) and return its ACTUAL structure so you can analyze it before writing something similar: page title, meta description, the H1/H2/H3 heading outline, word count, image count, and the readable body text (trimmed). USE THIS whenever the user gives a URL to 'follow the structure of' / 'write like this post' - never claim you matched a reference you haven't actually read. Returns JSON. Only http/https public pages; not for the user's own WP admin (use get_post for that).

Schedule a post to auto-publish at a future time. `datetime_iso` must be site-local time in ISO format, e.g. '2026-07-15T09:00:00'. Sets status=future.

Search across all posts & pages by keyword. Returns id, title, url, type. Great for finding internal-linking targets before editing an article.

Edit an item of ANY custom post type. Only non-empty fields change. `terms` = 'taxonomy:id[,id]' to (re)assign custom-taxonomy terms (find IDs with list_taxonomies). Pass status='publish' to make a draft live. `slug` changes the URL and 301-redirects the old one automatically (needs the Redirection plugin). `author_id` sets the author (use list_authors for the id) - this is how you set the author on a custom post type, since set_post_author only works on normal posts.

Edit a page by ID. Only non-empty fields change. content replaces whole body (HTML). `slug` changes the page's URL - the OLD url is automatically 301-redirected to the new one (needs the Redirection plugin) so rankings and links are preserved.

Edit a post by ID - only non-empty fields change, content REPLACES the whole HTML body, category_ids/tag_ids are COMMA-SEPARATED ID STRINGS like '3,7' (never arrays), status='publish' makes a draft live; use update_post_body_keep_schema instead if the post has JSON-LD schema you must preserve. `slug` changes the post's URL - the OLD url is automatically 301-redirected to the new one (needs the Redirection plugin) so rankings and links are preserved.

Generate a HIGH-QUALITY, full-length article BODY in clean HTML using the premium writing model (Gemini 3.1 Pro), so the expensive model is used ONLY for the final writing while your cheaper model does all the research/planning. Use this to WRITE the article once you have the plan: pass a detailed `brief` (topic, angle, the exact section outline / headings to cover, any facts or reference-structure notes from read_reference_page), the `target_words` to hit, `section_count` (e.g. 28 to match a reference), optional `tone` and comma-separated `keywords`. Returns publish-ready HTML (h2/h3, paragraphs, lists) - NO title tag, NO surrounding html/body. Then pass the returned HTML to create_post / p

Compress the media library's JPEG/PNG images (and optionally convert to WebP) to speed up the site. `quality` 40-95 (default 80). to_webp=True writes .webp copies. apply=False previews sizes; apply=True actually processes (backs up originals first when compressing in place). To keep each run fast and safe, apply processes at most 25 images per call - since it always starts from the still-unoptimized images, just call it again to continue the next batch. Returns bytes saved. Requires wptaskify Studio.

Permanently delete a media item by id.

Find media (images) NOT used as any post's featured image and NOT referenced in any post body. apply=False previews; apply=True deletes them (frees space). Be careful - deletion is permanent. Scans up to `limit` media items.

Find media items with empty alt text. With apply=False (default) just reports them. With apply=True, sets alt text from each image's title (cleaned). Good for accessibility + image SEO.

AI alt-text generator: for images with EMPTY alt text ONLY (it never touches images that already have alt - safe to re-run), use Gemini VISION to look at each image and write a proper descriptive alt text (better than the title-based fix_missing_alt_text).

Generate a REALISTIC PHOTO featured image with Google Gemini and attach it to an EXISTING post (the post_id must already exist - create the post first). If `prompt` is empty, an automatic prompt is built from the post's ACTUAL title/topic (so a tulip article gets a tulip image, a dirt-bike article gets a dirt-bike image). The image is uploaded to the WP media library and (by default) set as the post's featured image. Returns the new media id + url.

Generate a realistic image with Gemini from a free-text prompt and upload it to the media library (NOT attached to any post). Returns media id + url. Use when you just need an image in the library.

Generate an image with Gemini and INSERT it inside a post's body as an <img>. If `after_text` is given, the image is placed right after the first occurrence of that text; otherwise appended before the schema block. Preserves schema.

List media library items (images). Returns id, title, source_url, mime_type.

Download a file from a URL straight onto the server.

Set a post's featured image (thumbnail) to an existing media item id.

Set the site's FAVICON (browser-tab / bookmark icon, aka Site Icon). `image` is either a public image URL (it gets uploaded) or an existing media id. WordPress wants a square image at least 512x512. Works on any theme (it's a core setting).

Set the site's LOGO (the custom logo shown in the header by themes that support it). `image` is a public image URL (uploaded) or an existing media id. Uses the core Site Logo setting (works on block themes and classic themes with logo support). If the theme doesn't support a logo, it's stored but may not display - use set_site_icon for the favicon, which every theme shows.

Edit a media item's metadata: title, caption, description, or alt text. Only non-empty fields change. (To just set alt text on many images use fix_missing_alt_text.)

Download an image from a public URL and upload it to the WP media library. Returns the new media id (use it as featured_media_id when creating/updating a post).

Scan published posts and report which are MISSING their SEO meta title, description, or focus keyword. Great for finding posts to optimize.

Generate SEO meta with AI: meta TITLE (<=60 chars), meta DESCRIPTION (<=155 chars) and a FOCUS KEYWORD - not just the description. Set post_type='product' to run it on WooCommerce PRODUCTS (a product is a post), or any CPT. `fields` = comma list of which to fill (default all three; e.g. 'title,description' to skip the keyword). apply=False previews; apply=True saves via the SEO backend (update_post_seo under the hood). Only items with real prose are processed.

Delete ALL leftover _aioseo_* meta from every post (cleans old All in One SEO data so only your theme's _cwg_seo_* SEO fields remain). Pass confirm=True to run. This does NOT touch your _cwg_seo_* fields or post content. Run scan_aioseo_leftovers first to see what will be removed. IMPORTANT: if you still need the AIOSEO data, run migrate_aioseo_meta(apply=true) BEFORE this - cleaning is irreversible.

Compare ONE of your posts/pages against a competitor URL that outranks it - a content-depth scorecard so you can make YOUR page demonstrably more complete (the #1 lever on actual ranking). Fetches the competitor's public page and compares, side by side: word count, heading count, does it have an FAQ, images, schema (Article/FAQ), lists/tables, and the subtopics (H2/H3 headings) THEY cover that you don't. Pure code, no paid data - it can't show their backlinks/traffic/ keywords, and says so. Use this before refreshing a post so you know exactly what to add.

Create a redirect (e.g. fix a 404). source_url = old path like '/old-page/'; target_url = new path or full URL. 301 = permanent (default).

Delete a redirect by its ID (find it with list_redirects). Removes it from the Redirection plugin permanently. Use for duplicate/stale redirects.

Get or set the site's robots.txt (served via WordPress). Pass get_only=True to just read it. Setting it controls what search engines can crawl - be careful not to block the whole site.

Find & replace text across an SEO meta field on many posts. `field` is one of: meta_title, meta_description, focus_keyword. apply=False previews matches; apply=True writes the changes. Great for bulk rebrands/typos in SEO fields.

Find posts that are LOSING search traffic/position so you can refresh them - the single highest-ROI SEO activity. If Google Search Console is connected, it compares the last `days` to the previous `days` and flags pages whose clicks or position dropped. If GSC is NOT connected, it falls back to a content-age heuristic (older posts that are also thin get flagged as refresh candidates). Returns a ranked 'refresh these first' list; then use improve_old_post to fix them.

Find posts with identical or near-identical titles (possible duplicate/competing content). Returns groups of posts sharing a title.

Scan posts and flag THIN content (fewer than `min_words` words of visible text, schema/HTML stripped). Thin pages hurt SEO. Returns id, title, word_count, url sorted shortest first.

Return the site's XML sitemap URL (wptaskify serves one automatically). NOTE: Google (June 2023) and Bing both RETIRED the old sitemap-ping URLs, so there is no reliable ping endpoint anymore - reliable resubmission is done via Google Search Console (Site Kit). This confirms the sitemap and, if the old ping endpoints happen to still respond, reports that too.

Low-level: read ANY custom meta field on a post by exact key name. Use the diagnostic /wg-seo/v1/keys endpoint names. For advanced/edge cases.

Read the SEO meta title + description for an AUTHOR archive page. WordPress makes the author archive but gives it no SEO meta - this does. Requires the wptaskify plugin. Use list_authors / list_users to find user IDs.

Extract ONLY the JSON-LD schema block(s) from a post's content. Use this to read/inspect a post's structured data without the full article. For a CUSTOM post type (e.g. 'service'), pass post_type.

Read a post's SEO meta fields (meta title, description, focus keyword, keywords). Works on ANY site - auto-detects wptaskify, Yoast, Rank Math, or the wg-seo bridge. Returns current values + length hints (+ score if the wptaskify plugin is installed).

Read the SITE-WIDE SEO settings (wptaskify plugin): Knowledge Graph identity (Organization vs Person, name, logo, sameAs social profiles, knowsAbout topics), homepage SEO title/description, and the default social (OG) image. Requires the wptaskify plugin. Use update_seo_settings to change them.

No parameters.

Read the SEO meta title + description for a TERM archive (category, tag, or custom taxonomy term). WordPress makes these archive pages but gives them no SEO meta - this does. Requires the wptaskify plugin. Use list_categories / list_tags to find term IDs.

FREE multi-source keyword research: real GOOGLE + BING Autocomplete suggestions, grouped into questions / comparisons / buyer-intent / modifiers, then (use_ai=True) blended by AI into a prioritized topic plan. No paid SEO tool needed. `seed` = a topic/keyword (e.g. 'commuter e-bike'); `region` = 2-letter country (us, in, uk...). These are real search suggestions, NOT exact volumes - label any volume as an estimate.

List recent 404 errors visitors hit (from Redirection plugin's 404 log). Great for finding broken URLs to redirect. Returns url + hit count.

List existing redirects (Redirection plugin) with their ID, source URL, target, and hit count. Use `search` to filter by URL. Get the ID here, then delete_redirect to remove a duplicate/unwanted one.

One-shot migration: copy old All-in-One-SEO (_aioseo_*) meta into your theme's SEO fields (_cwg_seo_*), for ALL posts + pages in a SINGLE server-side pass (no per-post loop). SAFE: only fills EMPTY target fields (never overwrites SEO you set manually), resolves AIOSEO template tags (#post_title, #site_title, #separator_sa) into real text, and flags anything with an unresolved tag.

Ask search engines to re-crawl. IMPORTANT: Google and Bing RETIRED their sitemap-ping endpoints in 2023, so there is no working "ping" anymore. The reliable way to get new content crawled is Google Search Console (Site Kit) - submit your sitemap once and use "Request indexing" for a specific URL. This tool now returns those instructions instead of hitting dead endpoints.

Count leftover _aioseo_* meta data still stored in the database (from the old All in One SEO plugin). These can conflict with your theme's _cwg_seo_* fields.

Full on-page SEO audit of ONE post in a single report: title length, excerpt/meta length, word count, heading structure (H1-H6), internal vs external link counts, images & missing alt, schema present?, and (if given) focus_keyword usage in title/first paragraph/density. Returns issues + stats. For a CUSTOM post type (e.g. 'service'), pass post_type.

Detect which SEO system this site uses (wptaskify, Yoast, Rank Math, wg-seo bridge, or none). Tells you whether SEO meta tools will work here and suggests installing the free wptaskify plugin if nothing is detected.

Low-level: set ANY custom meta field on a post by exact key name. Use carefully.

FREE trending topics from Google Trends (today's rising searches) for a region - useful for timely article ideas. `region` = 2-letter country (us, in, uk...). `niche` (optional) = filter/relate results to your site's topic by also pulling autocomplete for it. Returns current trending search terms.

Set the SEO meta title and/or description for an AUTHOR archive page. Only non-empty fields change. Requires the wptaskify plugin.

Replace the article body but PRESERVE the existing JSON-LD schema block. Pass `new_content` as the new article HTML WITHOUT schema; the post's current schema <script> is automatically re-appended so structured data is not lost. Use this for normal content edits on this site. For a CUSTOM post type (e.g. 'service'), pass post_type.

Replace ONLY the JSON-LD schema block in a post, leaving the article body untouched. `new_schema_script` must be a full <script type="application/ld+json"> ... </script> tag. If the post has no schema yet, it is appended at the end. For a CUSTOM post type (e.g. 'service'), pass post_type.

Set a post's SEO meta_title, meta_description and focus_keyword (keywords = a COMMA-SEPARATED string of secondary keywords, not an array); only non-empty values are written. Works on ANY site - auto-detects wptaskify, Yoast, Rank Math, or the wg-seo bridge, and writes to the right fields so the rendered <title>, meta description, Open Graph & Twitter tags update.

Update SITE-WIDE SEO settings (wptaskify plugin). Only non-empty fields are changed (partial update). Fields: - entity_type: 'Organization' or 'Person' (who publishes this site) - entity_name: brand/person name (blank = use site title) - same_as: social profile URLs, one per line (Facebook, X, LinkedIn, etc.) - knows_about: topics this site is an authority on, one per line - home_title / home_desc: homepage SEO title & meta description - logo_id / default_og_id: media library attachment IDs (upload first) This powers the site's Organization/Person schema + default OG image. Requires the wptaskify plugin (admin capability).

Set the SEO meta title and/or description for a TERM archive (category, tag, custom taxonomy term). Only non-empty fields change. Improves how the archive page ranks and appears in search. Requires the wptaskify plugin.

Validate the JSON-LD structured data (schema) in a post - or scan many posts if post_id=0. Checks each <script type="application/ld+json"> block parses as valid JSON and has @context and @type. Reports posts with missing or broken schema. Good for rich-results health.

Fetch the LIVE rendered page (cache-bypassed) and extract the ACTUAL meta tags Google sees: title, meta description, canonical, OG title/desc, Twitter, and which plugin/theme emitted them (looks for the <!-- cwg-seo --> marker). Use this to confirm an SEO edit really rendered. Pass post_id OR a full url.

The AI SEO Score - a unified, AI-era SEO scorecard for the whole site across 5 categories (each 0-100) plus an overall: On-Page, Technical, AEO (answer- engine readiness), GEO (AI-citation readiness), and Authority (E-E-A-T). Every score is MEASURED from real content/site signals (no external APIs). Also returns concrete issue COUNTS (missing meta, missing alt, no schema, orphan pages, thin content, broken structure) so the dashboard can show 'Fix with Claude' actions. This is what powers the dashboard's AI SEO widget.

Get or set the site's /llms.txt - the AI-friendly index (like robots.txt but for LLMs / AI answer engines). It tells ChatGPT, Perplexity, Gemini etc. what the site is about and which pages matter, so they understand and CITE it better (GEO/AEO). get_only=True reads current state; passing `contents` sets a custom llms.txt (empty string = back to auto-generated from the site). Great to write a curated, well-described index of your best content. Requires wptaskify Studio.

GEO / AEO audit - how CITATION-READY a post is for AI answer engines (ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude). Scores 8 dimensions 0-100 (clear definitions, quotable statements, factual density, source citations, structured Q&A format, authority/E-E-A-T, freshness, structure clarity), gives an overall GEO score, and runs the ANSWER-BLOCK TEST against your `target_queries` (comma-separated) - for each query, is there a standalone 25-50 word answer a machine could quote? Returns measured scores + specific, actionable fixes. All numbers here are MEASURED from the content (not guessed); anything inferred is labeled. For a CUSTOM post type (e.g. 'service'), pass post_type.

Make a post AI-CITATION-READY. Two ways to use it:

Set a post's ANSWER-ENGINE (AEO/GEO) fields - the stuff that gets a page quoted and cited by ChatGPT, Perplexity, Google AI Overviews, Gemini & Claude. WordPress and most SEO plugins have none of this. Only non-empty fields change. Requires the wptaskify plugin. Fields: - quick_answer: a standalone 25-50 word direct answer (rendered at the top of the post + marked speakable). This is the #1 GEO lever. - key_takeaways: TL;DR bullets, ONE PER LINE (rendered as a list + schema abstract) - reviewed_by: E-E-A-T reviewer name (adds reviewedBy to Article schema) - last_reviewed: intentional review date YYYY-MM-DD (separate from auto-modified) - speakable: '1' to mark the title + quick answer as voi

STEP 2 of bulk internal linking: EXECUTE a plan from `plan_internal_links`, one BATCH at a time - no re-scanning, no per-post confirmation.

Get a temporary download URL for a backup file so the user can pull the backup OFF their site (e.g. to move it to another host). The big file is downloaded directly from the site in the browser - it does NOT come through this connector. what='files' for the code/media zip, what='db' for the database.sql (only exists if the backup was made with include_db=True). The link expires after ttl seconds (default 1 hour). Give the user the returned download_url to open in their browser. Requires wptaskify Studio.

BULK internal linking across the WHOLE site in ONE call - no need to go post-by-post. For every published post it finds natural anchor phrases that point to OTHER posts and turns the first mention of each into an internal link.

Check links inside a post (or scan many posts if post_id=0) and report any that return 404 / errors. Checks both internal and external links. For a full scan, raise `limit` carefully (each link is fetched). Returns broken links per post.

Find ORPHAN content: published posts/pages that NO other post links to internally. Orphans are hard for users and search engines to discover. Returns the orphaned items so you can add internal links to them (e.g. with bulk_internal_links). Scans up to `limit` posts.

Find posts most related to a given post (by shared title keywords). Useful for adding 'Related articles' internal links. Returns title + url.

STEP 1 of bulk internal linking: scan the WHOLE site ONCE and produce a COMPLETE plan of every internal link to add - WITHOUT changing anything.

Given a topic/keyword (or a draft article's subject), find the most relevant EXISTING posts to link to internally. Returns title + url + why (which topic words matched). Results are RANKED by how many topic keywords each title shares, so the most on-topic targets come first. Use this before writing/editing so you can insert internal links to these targets.

Set the permalink structure (e.g. '/%postname%/', '/%year%/%monthnum%/%postname%/'). Then flushes rewrite rules. Changing this can affect existing URLs - warn the user. Requires wptaskify Studio (uses option write + flush).

Add a custom link item to a navigation menu. Returns the new item id.

Add a block widget to a widget area (sidebar). `block_html` is Gutenberg block markup, e.g. '<!-- wp:paragraph --><p>Hello</p><!-- /wp:paragraph -->' or a heading/list block. Find sidebar_id via list_widgets. Great for footer text, a CTA, or an ad block.

Assign categories and/or tags to MANY posts at once. `post_ids`, `category_ids`, `tag_ids` are comma-separated id lists. By default terms are ADDED; set replace=True to replace existing ones. Returns per-post result.

Move ALL posts from one category to another (merge/cleanup). Every post in `from_category_id` gets `to_category_id` added and the old one removed. Returns how many posts were moved.

Create a new category. Returns its id.

Create a new (empty) navigation menu. Optionally assign it to a theme location (e.g. 'primary' - find valid locations in list_menus output). Then use add_menu_item to fill it. Returns the new menu id.

Create a new tag. Returns its id.

Delete a navigation menu (its items are removed too; the menu location becomes empty).

Remove an item from a navigation menu.

Delete a taxonomy term (category, tag, or custom-taxonomy term). Posts keep existing but lose this term. `taxonomy` = REST base ('categories'/'tags'/custom rest_base). Terms are deleted permanently (WP has no term trash).

Remove a widget from its widget area (by widget id from list_widgets).

List all categories with id, name, count, slug. Use IDs when assigning to posts.

List items in a menu (id, title, url, order, parent). Use to see/plan menu edits.

List navigation menus (id, name) and their locations.

List tags with id, name, count, slug. Optionally filter by `search`.

List the taxonomies (categories/tags AND custom ones like 'service_hub') on the site, optionally filtered to a post_type, with their REST base and (if include_terms) their terms + term IDs. Use this to find the term IDs to assign when creating a custom post-type item.

List the site's widgets and widget areas (sidebars). Returns each sidebar's id/name and the widgets in it. Many classic themes use widgets for footers/sidebars.

Edit an existing menu item: rename, change its URL, re-nest under a parent, or reorder (menu_order). Only non-empty/>=0 fields change. Use to reorder or restructure a menu without deleting + re-adding items.

Rename / re-describe / re-slug / re-parent a taxonomy term. `taxonomy` is the REST base: 'categories', 'tags', or a custom taxonomy's rest_base (find via list_taxonomies). Only non-empty fields change. parent_id=-1 leaves parent unchanged (categories only).

Change a user's role (subscriber/contributor/author/editor/administrator).

Post a comment on a post (or REPLY to another comment via parent_id). Use to answer reader comments as the site. If author_name/email are omitted the comment is authored by the connected account. New comments are auto-approved (posted by an admin).

Create a new WordPress user. role is one of: subscriber, contributor, author, editor, administrator. Returns the new user id.

Delete a user. reassign_to = the user id to reassign their posts to (recommended so content isn't lost). If reassign_to=0, their content is deleted too - be careful.

Get one user's profile (name, email, bio, url, roles). user_id='me' = the connected account. Use before update_user to see current values.

List all authors/users (id, name, role). Use the id with set_post_author.

List comments. status = hold (pending), approve, spam, trash, all. Returns id, author, content snippet, post id, status.

List site users with id, name, email (if visible), and roles.

Moderate a comment. action = approve, hold, spam, trash, delete.

Set/change the author of a post (use list_authors to get the id). Good for E-E-A-T - assign the right expert to each article. For a CUSTOM post type (e.g. 'service'), pass post_type so it targets the right endpoint - otherwise it only works on normal posts.

Edit the CONTENT of an existing comment (fix a typo, redact something). To change a comment's status (approve/spam/trash) use moderate_comment instead.

Edit a user's profile: display name, email, first/last name, bio (description), website url, or password. Only non-empty fields change. Use get_user first to see current values.

Activate an installed plugin by its file (e.g. 'akismet/akismet.php' - get it from list_plugins). Requires wptaskify Studio.

Activate a theme by its slug (folder name) - makes it LIVE for all visitors.

Create a NEW WordPress plugin (folder + main file with a valid header). `slug` = folder/file name (lowercase-dashes). `code` = optional PHP body AFTER the header (do NOT include <?php or the header - those are added). Safe: a new plugin can't break existing code. Activate it separately if needed. Requires wptaskify Studio.

Create a NEW WordPress theme (style.css + index.php + functions.php). `slug` = folder name. `style_css` = CSS (a Theme Name header is added if missing). `index_php`/`functions_php` = optional PHP (lint-checked). Safe: creating a theme doesn't affect the active one until you activate it with activate_theme. Requires wptaskify Studio.

Deactivate an active plugin by its file (from list_plugins). wptaskify cannot deactivate itself. This can change site behaviour - WARN the user and confirm first. Requires wptaskify Studio.

Get or set the root .htaccess file. RISKY - a wrong rule can break the whole site (500 error). The plugin backs up the old file first. Always read it first (get_only=True), WARN the user, and confirm before writing. Requires wptaskify Studio.

Get the current site-wide custom CSS set via wptaskify Studio.

Install a plugin from the WordPress.org repository by its slug (e.g. 'wordpress-seo', 'contact-form-7'). Set activate=True to also activate it. Requires wptaskify Studio.

List all installed plugins with name, version, file, and active/inactive state. Requires wptaskify Studio.

List files/folders inside a theme, plugin or the uploads dir. `path` must start with 'themes/', 'plugins/' or 'uploads/' (e.g. 'themes/blocksy' or 'plugins/wp-pilot-seo'). Use this to explore before reading/editing files. Requires wptaskify Studio.

List all installed themes with name, version, and which one is ACTIVE. Use this before activating/previewing/editing a theme. Requires wptaskify Studio.

Get a SAFE PREVIEW link for a theme WITHOUT activating it - only the logged-in admin sees the preview; visitors keep seeing the current live theme. This is the 'staging' step: create_theme -> preview_theme -> (looks good?) -> activate_theme. ALWAYS preview a new theme and let the user confirm it looks right before activating. Requires wptaskify Studio.

Read a theme/plugin/uploads file's contents. `path` starts with 'themes/', 'plugins/' or 'uploads/' (e.g. 'themes/blocksy/functions.php'). Requires wptaskify Studio. Read BEFORE editing so you preserve existing code.

Create a new plugin on the filesystem, with a valid header.

Create a new theme on the filesystem, with the files WordPress requires.

Run a read-only SQL query against the site's own database.

Delete a file, or an empty folder.

Disable ONE plugin by renaming its folder - works on a site that will not load.

Force WordPress back to a default theme, by renaming the active theme's folder.

Change part of a file without sending the whole thing back.

Undo server_disable_plugin - rename the folder back.

Install a plugin from a .zip URL by unpacking it into wp-content/plugins.

List a folder inside the site - name, type, size, modified time.

Create a folder, and any missing folders above it.

PHP version, memory limit, execution time and extensions, as the SITE sees them.

Read a file from the site's filesystem.

Rename anything inside the site - the general form of the disable tools.

Search the site's files - where is this function, who writes this option.

Write or replace a file on the site. PHP is syntax-checked FIRST.

Set the site-wide custom CSS (applies to the WHOLE site, every theme). This is the SAFEST way to restyle a site - colors, fonts, spacing, layout tweaks - because CSS can never crash PHP. REPLACES the current custom CSS; use get_custom_css first if you want to append. Requires wptaskify Studio.

WRITE / EDIT / SAVE / UPDATE a theme or plugin file - create or overwrite. Use this to edit theme PHP files (functions.php, front-page.php, index.php, header.php, page templates), style.css, or any theme/plugin/uploads file. `path` starts with 'themes/', 'plugins/' or 'uploads/' (e.g. 'themes/mytheme/front-page.php'). This is the theme/ plugin FILE WRITE tool. The Studio plugin AUTOMATICALLY backs up any existing file and, for .php files, PHP-lints `contents` first - if the PHP is invalid the write is REFUSED (so a bad edit can't take the site down).

Check whether a queued approval (from request_approval) has been decided by the user. Returns status: 'pending', 'approved', or 'rejected'. Only perform the action once status is 'approved'; if 'rejected', do not do it.

Report site health: PHP & WP version, active theme, plugin counts, DB size, HTTPS, debug mode, memory limit. Good for a quick diagnostic. Requires wptaskify Studio.

Take a FULL backup of the site. Always includes code (all themes + plugins + mu-plugins). Set include_uploads=True to also back up the media library (can be large). Set include_db=True for a MIGRATION-GRADE backup - a complete SQL dump of EVERY database table (posts, users, options, WooCommerce, everything), built in resumable chunks so it never times out. With include_db the backup can rebuild the whole site on a new host. Returns a backup_id to use with restore_site_backup. Do this BEFORE big/risky changes or a migration. Requires wptaskify Studio.

Get the recent wptaskify activity log (what the AI changed on this site). Requires wptaskify Studio.

List the automatic file backups wptaskify Studio has made (before each edit). Useful to reassure the user their originals are saved. Requires wptaskify Studio.

Queue a genuinely DESTRUCTIVE / hard-to-undo action for the site owner to approve, instead of doing it right away. Only a SMALL set of actions ever needs this.

Record the user's answer to a pending approval that they gave IN THIS CHAT.

Roll a post back to a previous revision (undo). Fetches that revision's content/title and writes it back onto the live post.

Restore a full-site backup by its backup_id (from full_site_backup). This overwrites theme/plugin/mu-plugin files and restores the active theme/plugins. Set restore_db=True to ALSO import the database.sql from a migration-grade backup - this OVERWRITES the site's entire database (all posts, users, options). VERY RISKY - only for migration/disaster recovery. WARN the user and confirm first. Requires wptaskify Studio.

Instantly switch back to the theme that was active BEFORE the last activate_theme. Use if a newly activated theme looks broken or wrong on the live site. Requires wptaskify Studio.

List the backups taken of one file, newest first.

When a site is down and you do not know why - run this FIRST.

Disable EVERY plugin at once, by renaming the plugins folder.

Read the PHP error log - the real reason a site is broken.

Reset file and folder permissions to what WordPress expects: 644 and 755.

Undo server_disable_all_plugins.

Turn WordPress debug logging on or off in wp-config.php.

Is server (SFTP/cPanel) access connected for this site, and what can it reach?

Put a file back to how it was before the last write.

Add a note to an order. customer_note=True emails it to the customer; False = private (staff-only) note.

Generate descriptions for many products at once. only_missing=True (default) targets products with an EMPTY description. apply=False previews the list that would change; apply=True writes them. Great for filling out an imported catalog. To protect quality, apply runs in batches of at most 25 per call - use `offset` to continue (offset=25, then 50...) and review between batches.

Generate professional AI product images for MANY WooCommerce products at once. Each image is built from that product's own name + category + description (a rich, specific prompt), so the catalog looks like a real photo set - NOT generic. Quality is identical to the single generate_product_image (same AI provider, same prompt quality - no shortcuts for being bulk).

Generate SEO meta titles + descriptions for WooCommerce products (products are posts, so this uses the SEO backend). apply=False previews which products would change; apply=True writes meta via the SEO plugin. To protect quality, apply runs in batches of at most 25 per call - use `offset` to continue (offset=25, then 50...) and review between batches.

Write SEO alt text for WooCommerce PRODUCT images - YOU (the connected AI) look at the images and write the alt text, so it's specific and high quality (not generic).

Create a WooCommerce coupon. discount_type = percent / fixed_cart / fixed_product. `amount` is '10' (=10% or a fixed amount). Optional expiry (YYYY-MM-DD), usage_limit, min_amount (minimum cart total), and free_shipping.

Create a WooCommerce product on the connected store and return its id. Status defaults to 'draft' so it's safe. Covers every common WooCommerce product option: * basics: name, regular_price/sale_price ('19.99'), description, short_description, sku, product_type (simple|grouped|external|variable), status, featured * inventory: stock_quantity (>=0 turns on manage_stock), stock_status (instock|outofstock|onbackorder) * taxonomy/media: category_ids + tags (comma names or ids), image_urls (comma, public URLs) * shipping: weight, length/width/height, shipping_class * tax: tax_status (taxable|shipping|none), tax_class * type flags: virtual (no shipping), downloadable ("true" or a JSON list of {"nam

Create a WooCommerce product category. Returns its id.

Add ONE variation to a VARIABLE WooCommerce product. First create the parent with product_type='variable' and attributes that have variation=true, then call this once per variation (e.g. Red/S, Red/M, Blue/S...). * attributes = JSON mapping the varying attributes to THIS variation's values, e.g. {"Color":"Red","Size":"S"} (names must match the parent's variation attributes). * regular_price/sale_price ('19.99'), sku, stock_quantity (>=0 = managed), stock_status * weight, image_url (one public URL), description, virtual, downloadable, meta_data (JSON).

Delete a WooCommerce coupon permanently.

Delete a WooCommerce product. Default = move to Trash. permanent=True deletes forever.

Delete a WooCommerce product category (products keep existing, lose this category).

Generate a compelling product description + short description with AI, from the product's name, attributes and any existing detail. `tone` guides the style (persuasive/technical/ friendly). `keywords` are SEO terms to work in naturally. apply=False PREVIEWS the copy; apply=True saves it. For product SEO meta (title/description), call update_post_seo with this product id afterwards - a product IS a post, so no post_type argument is needed.

Generate a product image with AI and attach it to a WooCommerce product. If `prompt` is empty, a rich one is built from the product name, category and description (better quality). set_as_featured=True makes it the main image.

Get a WooCommerce order's full detail: line items, customer, billing/shipping, totals, payment method and status.

Get a WooCommerce product's full detail: description, short description, price, stock, SKU, categories, attributes and images.

Build a polished Amazon product-review CARD and insert it into a post - the main monetization tool. It fetches the product (by ASIN, or the first match for `keywords`), uploads up to `real_images` real Amazon image(s), optionally generates 1 extra polished image with Gemini, and inserts a card (image(s) + title + price + features + 'View on Amazon' affiliate button + disclosure) after `after_text` (or before the schema block). Preserves the post's JSON-LD. Needs Creators API credentials for real data/images; without them it still builds a card with a Gemini image + a tagged search link.

List WooCommerce discount coupons (code, type, amount, usage, expiry).

List WooCommerce orders (id, status, total, customer, date, items). Filter by `status` (pending/processing/completed/cancelled/refunded/any), `customer` (a customer id), or `after` (YYYY-MM-DD, orders created after this date).

List WooCommerce product categories (id, name, slug, count, parent).

List all variations of a variable WooCommerce product (id, sku, price, stock, and the attribute values each variation stands for). Read-only.

List WooCommerce products (id, name, sku, price, stock, status). Filter by `search`, `status` (publish/draft/any) or `category` (a category id). Use before editing.

Put a product on sale. sale_price is a string ('14.99'). Optional date_from/date_to (YYYY-MM-DD) schedule the sale window. Pass sale_price='' via update_product to end a sale.

Quick store performance summary: total sales, order count and top sellers for a period (week / month / year). A fast 'how is my store doing' snapshot.

Edit a WooCommerce coupon: change its code, discount_type (percent/fixed_cart/ fixed_product), amount, expiry (YYYY-MM-DD), usage_limit, min_amount, or free_shipping ('yes'/'no'). Only provided fields change.

Change a WooCommerce order's status: pending, processing, on-hold, completed, cancelled, refunded, or failed. Optional `note` is added to the order as a private note.

Edit a WooCommerce product. Only non-empty fields change (nothing you leave blank is touched). Prices are strings ('19.99'). Pass status='publish' to make a draft live. Same rich fields as create_product: * category_ids / tags (comma ids or names), stock_status (instock|outofstock|onbackorder) * weight, length/width/height, shipping_class, tax_status, tax_class * virtual / featured / reviews_allowed = "true" or "false" (blank = leave unchanged) * downloadable = "true" or a JSON list of {"name","file"} * cross_sell_ids / upsell_ids (comma ids), purchase_note, menu_order * attributes = JSON list (see create_product), meta_data = JSON object/list to write ANY custom field / ACF / plugin meta. T

Edit a WooCommerce product category: rename, re-describe, re-parent or re-slug. Only non-empty fields change. parent_id=-1 leaves the parent unchanged.

Quick inventory edit for a product. stock_quantity sets a managed count; stock_status = instock / outofstock / onbackorder (for products without a count).

Create a WooCommerce product tag. Returns its id.

Create a variation of a variable product. attributes_json maps each attribute to a value, e.g. '[{"name":"Size","option":"M"},{"name":"Color","option":"Red"}]'. regular_price is a string. Set the product's attributes first with wc_set_product_attributes.

Get a WooCommerce customer's detail incl. billing/shipping.

List WooCommerce customers (id, name, email, orders, total spent). `search` matches name/email.

List WooCommerce product reviews (reviewer, rating, status, text). Optionally filter to one product_id.

List WooCommerce product tags (id, name, count).

List a variable product's variations (id, attributes, price, stock).

Moderate a product review. status = approved / hold / spam / trash.

Issue a refund on a WooCommerce order. If `amount` is empty, refunds the full order total. `reason` is an optional note. This moves real money - use carefully.

Set a product's attributes (needed for variable products). attributes_json is a JSON array like '[{"name":"Size","options":["S","M","L"],"variation":true,"visible":true}]'. After this, create variations with wc_create_variation.

Check whether WooCommerce is active on this site and get a quick store snapshot (currency, product count, order count). Use this FIRST before other wc_/product/order tools. Read-only.

Edit a WooCommerce customer's email or name. Only non-empty fields change.

Edit a WooCommerce order beyond status: customer_note, or billing/shipping address (JSON objects like '{"first_name":"Jo","address_1":"1 St","city":"NYC"}'). Only provided fields change.

Add an image to a product from a public URL.

Add a product to a manual collection.

Add a variant to a product (e.g. a new size/colour). option1 is the variant option value (e.g. 'Large'). price is a string.

Per-PRODUCT answer-engine readiness analyzer (the Shopify version of the plugin's AEO Analyzer). For each product, checks 5 signals AI answer engines (ChatGPT, Perplexity, Google AI) look for: a direct/quotable answer, an FAQ block, citation-ready facts+sources, a clear entity definition, and freshness. Returns a per-product grid + a summary of how many miss each signal. Pure code, zero AI cost.

Control which AI crawlers can read your store (like the WordPress AI Access Control). Writes a managed block into the theme's robots.txt.liquid that Disallows each bot you choose to BLOCK - without touching Shopify's own default rules. `block` = comma-separated bot names to block (e.g. 'GPTBot,CCBot'); anything not listed is ALLOWED. get_only=True just returns the current state.

AI-Readiness / AEO score - grades whether your store is VISIBLE to AI shopping engines (ChatGPT, Perplexity, Google AI Overviews, Gemini). This is the biggest 2026 gap and almost no free app scores it. All pure code (fetches robots.txt + your homepage + a product page, parses JSON-LD) - ZERO AI cost. Checks: (1) AI crawlers (GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended...) are NOT blocked in robots.txt - a store that followed old 'block the scrapers' advice is invisible to AI; (2) schema depth (Product, Offer, aggregateRating, Organization, BreadcrumbList, FAQPage) - AI answer engines cite structured data; (3) reviews/ social-proof signal present. Returns a 0-100 AI-readi

The Shopify AI SEO Score - a store-wide SEO scorecard across 4 categories (each 0-100) plus an overall, MEASURED from your real product data (no external APIs). Checks the things that actually move Shopify SEO + AI visibility: meta titles/descriptions, product description depth, image alt text, and AEO/GEO readiness (structured, answer-friendly content). Returns concrete issue COUNTS (missing meta, thin descriptions, images without alt) so the AI can offer 'fix this' actions. The Shopify equivalent of the WordPress ai_seo_score.

Audit EVERY collection for the SEO basics: page title, meta description and intro copy.

Scan the store for broken links / 404s - the classic feature other apps charge for, here FREE. Checks that each published product, collection, page and blog article's storefront URL returns 200 (not 404/dead). Reports the dead URLs so the merchant can fix or redirect them. Pure HTTP checks - no AI cost.

Generate AI descriptions for many Shopify products at once. only_missing=True targets products with an empty body_html. apply=False previews; apply=True writes them. Great for a freshly-imported catalog. To protect quality, apply is processed in batches of at most 25 per call - use `offset` to continue (offset=25, then 50...) and review between batches.

Generate professional AI product images for MANY Shopify products at once. Each image is built from that product's own title + type + description (a rich, specific prompt), so the catalog looks like a real photo set, NOT generic. Same AI quality as the single shopify_generate_product_image - no shortcuts for being bulk.

Write SEO alt text for Shopify PRODUCT images - YOU (the connected AI) look at the images and write the alt text, so it's specific and high quality (not generic).

Generate SEO meta titles + descriptions for many Shopify products at once (stored as Shopify global SEO metafields). apply=False previews; apply=True writes them. To protect quality, apply runs in batches of at most 25 per call - use `offset` to continue (offset=25, then 50...) and review between batches.

Publish a blog article to a Shopify blog (get blog_id from shopify_list_blogs). body_html is the article HTML. published=False saves it as a draft (safe). tags comma-separated.

Create a manual (custom) collection. Add products with shopify_add_to_collection.

Create a Shopify discount code. value_type = percentage or fixed_amount. `value` is the amount (e.g. '10' for 10% or a fixed amount; stored negative automatically). Optional starts_at/ends_at (YYYY-MM-DD) and usage_limit.

Create a static page (e.g. About, Shipping Policy). published=False saves a draft.

Create a Shopify product. `price` is a string ('19.99'). `tags` comma-separated. image_urls comma-separated public URLs. status defaults to draft (safe). Creates a single default variant with the given price/sku/inventory.

Create a URL redirect (e.g. from '/old-product' to '/products/new'). Great after deleting or renaming products/pages so old links don't 404.

Create a webhook (e.g. topic='orders/create', address='https://your-endpoint'). Lets you react to store events.

Declare wptaskify's product metafield DEFINITIONS on the store (one-time setup). This makes the AEO fields (quick answer, key takeaways, USP, best-for, trust points, FAQ, specs) show up as proper, editable fields in the Shopify product admin under the 'wptaskify' namespace - so a merchant can see and edit them by hand too, not only via the AI. Safe to run more than once: definitions that already exist are skipped. Run this once per store (or after adding a new field). Uses the Shopify GraphQL Admin API (metafieldDefinitionCreate).