Seovyn

Every Seovyn API endpoint (method, path, what it does), read from the server's own routes. Use it with seovyn_api_request for anything the named tools don't cover. Filter with `search` (a word in the path or description).

Call any Seovyn API endpoint as the user, in one organization (see seovyn_api_reference): { method, path like "/content/items/<itemId>", query?, body? }. GET runs at once; any other method returns a preview first and runs with confirm: true. The answer is the API's own status and data.

The autopilot. Seovyn starts supervised (you approve every article) and earns autopilot after a run of approvals; on autopilot it publishes good drafts on its own after a hold. action "status" (default): the mode, its last 20 decisions next to what autopilot would have done, recent autopilot events and the publishing pace; "set_pace" { maxArticlesPerDay: 1-36, or null for no pacing } — how many articles are written and go live a day; "turn_off": back to supervised (the approval streak starts again — needs confirm).

The brand profile every article is written to. action "get" (default); "update" with any of brandName, productOneLiner, audience, tone (practical | authoritative | friendly | technical), pointOfView (we | i | neutral), language, country (2 letters), authorName, ctaText, ctaUrl, bannedPhrases[], mustMention[]; "set_niche" { niche } (a sentence on what the site is about).

Live articles that need a decision (thin, overlapping or not earning). action "list" (default); "review" (look over the live articles now instead of waiting for the weekly pass); "decide" { itemId, decision: keep | unpublish | noindex } — unpublish takes the page down and noindex hides it from Google (Seovyn-hosted sites only): both need confirm.

Competitors' best pages by traffic, their sitemap sections, and whether this site covers each topic yet.

A new organization for another website: { name, website }. Or, for a business with no website, { name, noWebsite: true, businessName, description } and Seovyn builds and hosts a site for it. Seovyn then reads the site and plans its first keywords. It becomes the user's current organization in the dashboard too. Can't be undone from here: needs confirm.

Edit a draft before it is approved (drafts only): title, summary, body (Markdown), slug, metaTitle (≤70), metaDescription (≤170), tags (≤6). The quality checks run again on the saved version.

One article in full: the body (Markdown), title, slug, meta title and description, tags, keyword, the research it was planned against, its quality verdict, and where it stands (and what autopilot would do with it).

The site Seovyn hosts for a business without its own website. action "get" (default): the site, its address and pages; "change" { instruction } rewrites its pages as asked ("add a pricing section to the home page") and puts them live — needs confirm; "enquiries": messages visitors sent through its contact form.

Where articles are published (GitHub, WordPress, a CMS, a webhook or the Seovyn-hosted site) and each one's state, Google Search Console, and what the website is built on and the destination that fits it. Connecting one takes credentials: do it in the dashboard (Publishing), or with seovyn_api_request.

The keywords Seovyn writes for (each becomes an article): action "list" (default; with status open/written/skipped and where each came from), "add" { keyword, pageType? }, "remove" { keyword }.

What Seovyn knows first-hand about the business (facts read off the site, the owner's notes), which articles draw on. action "list" (default); "questions": what Seovyn would like the owner to answer; "answer" { questionId, answer, topic?, pageType? } (topic and pageType as the question gave them); "add_note" { text }; "remove" { factId }; "refresh": read the site again now.

Articles, newest first: status "pending" (the review queue: drafts waiting, being written, failed), "approved" (published or on their way), "rejected", or "all". Each with its keyword, quality score, where it went live and its search performance.

The organization's plan and billing. action "overview" (default): the plan, this period's usage, the next charge; "checkout" { plan: pro | scale }: a payment link for the user to open (nothing is charged until they pay); "change" { plan } for an organization already paying: shows what it charges now, and changes the plan with confirm.

Approved articles waiting their turn to go live (the publishing pace), with when each is due.

What the articles have earned: clicks, impressions and positions from Google Search Console, the best and rising articles, and how the site is doing since Seovyn started.

Decide on an article. decision "approve" (it publishes to the live site, now or at the publishing pace — needs confirm), "reject" (reason: inaccurate, off-brand, thin, duplicate, off-topic, bad-seo, other; note optional — Seovyn learns from it), "hold" (autopilot won't publish it on its own), "retry" (write a failed one again: uses one article of the month's allowance — needs confirm), "publish_refresh" (put the refreshed version waiting on a live article live — needs confirm) or "discard_refresh".

Rewrite one section of a draft to an instruction ("add a worked example", "shorter, no jargon"). index is the section's place (0 = the opening before the first heading) and heading its heading as it is now (null for the opening). Returns the new section and the whole body WITHOUT saving: save it with seovyn_edit_article. Uses one of the plan's monthly section rewrites: needs confirm.

Where Seovyn finds topics: action "list" (default), "add" { type: youtube | reddit | competitor, label, url, pageType? }, "discover" { sourceId } (look for new topics in it now), "remove" { sourceId } (needs confirm).

How an organization stands: counts (sources, drafts waiting, published, approved, rejected), the plan and this month's usage (articles, section rewrites), and every publishing integration's state.

Topic clusters (a pillar keyword and its supporting ones): action "list" (default), "plan" (plan more from the website now), "remove" { clusterId } (drops the cluster and its unwritten keywords — needs confirm).

Call this FIRST. Who this key acts as, whether it is an admin key, and the user's organizations (one per website: accountId, name, website, plan, supervised or autopilot, articles used this month). Tools that work inside an organization take `org` (its accountId, website or name); omitted, they use the user's current one.

Write an article now on a topic, against 1-4 competitor pages it should beat ({ topic, competitorUrls, pageType? }). It uses one article of the month's allowance and lands in the review queue as a draft (a minute or two). Needs confirm.