Leadstore.xyz

Mint a short-lived, single-use credential for connecting the Leadstore WordPress plugin (named create_, not get_ — this issues a new secret each call, it doesn't read one back). This only gets the user as far as having a token to paste — the plugin itself (installed and active on their real WordPress site) has to complete the registration call; there's no way to finish this from here. Rate-limited.

Remove the account's own saved API key for an assistant provider. Like saving, this is refused with 'provider_key_managed' for providers Leadstore supplies on the account's plan — there is no user-owned key to remove in that case.

Stop tracking a competitor.

Delete a keyword. Fails if a blog draft already covers it.

Delete a keyword group (does not delete the keywords in it).

Delete a past keyword-research run and its keywords.

Remove an on-page fix.

Delete a rank tracker and all of its tracked keywords and position history.

Remove a bookmarked competitor keyword.

Disconnect a site's Framer connection.

Disconnect a catalog integration.

Disconnect a site's Sanity connection.

Dismiss a content-gap topic so it stops showing up as open.

Get the real number of SERP API calls a run_rank_tracker call would make right now (one per tracked keyword) — check this before running a tracker with a lot of keywords.

Get how much of each metered feature is left this period (content_drafts, keyword_runs, visibility_runs, offsite_opportunities, engagement_drafts, page_optimizer_analyses, etc.), plus shape limits like audit_cooldown_seconds and max_keywords_per_run. Check this before suggesting any action that spends quota. Plan name, subscription status and billing details are deliberately NOT available here — if the user asks about their plan, pricing or subscription, tell them to open the Leadstore app rather than guessing.

Get attribution analytics for a site — which AI-driven touchpoints correlate with real traffic/conversions.

Get every check from a site's latest technical audit (SEO/AEO/GEO/LLMO/ASO), grouped by channel. Each failing check includes `detail` (what's wrong) and, for most failures, `ai_prompt` — a complete, ready-to-use prompt describing exactly how to fix it (file-level guidance, not just a description) that you can hand directly to a coding assistant, or use yourself to make the fix. Also includes `fix_impact`, the real point-gain from fixing each one.

List issues found by the latest audit's multi-page crawl (broken links, duplicate titles/meta descriptions, thin content, missing alt text, etc.) — a different, page-level layer from get_audit_checks. Optionally filter by severity (high|medium|low) or issue_type.

List every page from the latest audit's multi-page crawl, with per-page technical fields (status code, title, meta description, word count, image/link counts, indexability). Returns 404 if no audit has a crawl yet.

Check what's available before considering a new audit: the quota counters (including `shape_limits.audit_cooldown_seconds` and the `visibility_runs` counter) plus the audit score history (most recent entry's date tells you how long ago the last technical audit ran). Use this to tell the user whether they're likely still in cooldown or out of quota — running a new audit is something they do themselves in the Leadstore app; there's no trigger tool here.

Get the audit task tracker for a site: which failing checks are new, still open, resolved, or reopened since the previous audit. Read-only and derived — a task resolves itself once the underlying issue is fixed and a new audit is run; there's no separate mark-done action.

Get one blog draft's full content and metadata (list_blog_drafts may truncate the body). Two fields matter for improving a draft. `generation_plan` is what the draft committed to before it was written: the sections, which of the site's own data items (internal links, competitor keyword gaps, verified source URLs, real searched questions) were assigned to which section, the planned FAQ, and a stated reason for every item deliberately left out. `quality_report` is how it was measured against that: `scores` (structure_seo / grounding / trust), `grounding` (which committed items are actually missing from the body), and `findings` — fact-review problems, each carrying the verbatim quote from the draft it came from and the block it sits in. Fix the quoted passages and the missing committed items; that is what the score measures. `generation_plan.claim_evidence` holds the passage each cited page actually contained for a planned claim — check a figure in the draft against the wording it came from before treating it as sourced.

Get what AI engines are currently saying about the brand — the core "is this accurate/flattering" check, separate from the raw visibility score.

Get brand-monitoring results over time, not just the latest snapshot.

Get the trend of how often a site is being cited by AI engines over time.

Get which sources AI engines cite alongside this site when discussing it — the surrounding citation graph.

Get the sources AI engines draw from when they cite or discuss this brand.

Get recent competitor-visibility changes worth flagging — a competitor's mention share moving meaningfully since the previous audit run.

Get a tracked competitor's latest backlink profile snapshot. Returns 404 if none has been fetched yet — call refresh_competitor_backlinks first.

Get a tracked competitor's latest domain overview snapshot (organic traffic estimate, ranked-keyword count, domain rank) — real measured data, not just the name competitor discovery gives you. Returns 404 if none has been fetched yet — call refresh_competitor_domain_overview first.

Get the scheduled content plan items for a site, optionally within a date range.

Get a site's content-plan settings (mode, cadence, per-run limit, lookahead window, autopublish).

Get a site's Framer connection status.

Real session/user counts by acquisition channel (Organic Search, Paid Search, Direct, Referral, Social, Email, ...) from the connected GA4 property.

Real per-page key-event (conversion) counts from the connected GA4 property — previously this product's GA4 integration had no conversion/goal data at all, only pageviews/sessions/users.

Get a site's auto-generated llm-data.json file content.

Get a site's auto-generated llms.txt file content.

Get instances where an AI engine said something false or fabricated about the brand — the thing most worth flagging as "needs work."

Get hallucination findings over time, not just the latest snapshot.

Preview the other keywords clustered with a given keyword.