Automated SEO: Site Health, fix plan, Search Console, keywords and draft scoring for AI agents.
Scored across 46 tools
The 46 tools are mostly well-separated by resource and action, with clear namespacing for GA4/GSC and distinct roles for scan, issue, fix plan and remediation profile. A few boundaries are soft—get_project vs get_plan_status, get_site_health vs get_seo_progress vs get_results, and analyze_seo vs score_article—but the descriptions give enough guidance to choose correctly.
The set is consistently snake_case and predominantly verb_noun (get_*, list_*, apply_fix, start_scan). The ga4_* and gsc_* groups use a resource-prefix convention rather than verb-first naming, which is a minor deviation but still predictable and readable.
46 tools is heavy for a single MCP server, even one covering SEO, Search Console, GA4, AI visibility, content and site fixes. The breadth is real, but the count exceeds a comfortably scoped surface and increases selection cost for an agent.
Coverage is broad: account/projects, scans, health, issues, fix plans/runs, articles, clusters, keywords, GSC, GA4, AI visibility, channels and playbooks. Gaps remain around content creation/publishing and project/channel/keyword management, which are mostly read-only here.
Mark an issue thread as an intentional choice (write scope — ask the user first): it stops counting against the score and stays marked across scans. Typical uses: noindex on pagination, a deliberately short page, a redirect that must stay. Undo with unacknowledge_issue.
Inputs: issue_id (required), reason, project_id.
Run Seonix's deterministic SEO + readability + AI-answer (AEO) scoring on ANY draft you pass (title + HTML + meta + keyphrase). No AI credits are spent. Use it as a linter: fix the problems it lists, re-run until the score is ≥ 80, then publish.
Inputs: html (required), title (required), slug, language, +6 more.
Have Seonix apply fixes on the live site through the WordPress plugin (write scope — tell the user exactly which issue codes and get a yes first). Works for get_fix_plan items with fix_tier auto or ai_assist (wp_plugin_fixable); the site must have a connected WordPress channel. Starts a run (preview → Seonix suggestions → apply) and returns its id; poll get_fix_run until status is completed, partial or failed. Undo with rollback_fix.
Inputs: issue_codes (required), issue_ids, project_id.
Visits that came from AI assistants (ChatGPT, Perplexity, Gemini, Claude, Copilot…): totals, per assistant, and the pages they landed on.
Inputs: days, limit, project_id.
Landing pages by sessions for the window (top 50), with key events and the split of visits by source group (search, ai, social…). Pair with gsc_page_queries to see what a page ranks for.
Inputs: days, limit, project_id.
Every network that sent visitors in the window (Google, Bing, ChatGPT, Facebook, newsletters, referring sites…), grouped as search / ai / social / ads / email / referral / direct, with sessions and key events.
Inputs: days, limit, project_id.
Whether Google Analytics 4 is connected for the project and which property. Call before other ga4_* tools.
Inputs: project_id.
Google Analytics 4 totals for the window (sessions, users, pageviews, key events, bounce rate, average session), visits per channel with AI assistants as a channel of their own, and the landing pages behind the key events. include_daily adds sessions per day.
Inputs: days, project_id, include_daily.
The Seonix account behind this API key: email, every project with its plan, whether the MCP is available for it (paid plans only), AI/publish permissions and this month's article quota. Call this first.
AI Visibility: how often the brand appears in the answers of AI assistants (ChatGPT, Gemini, Claude, Google AI) for the project's tracked questions. Returns the overall score with its sample size, the per-engine breakdown, the weekly history, the competitors named, the top cited sources and the questions where the brand is absent. Checks run automatically once a week for projects on a paid plan; this tool only reads.
Inputs: project_id, include_answers.
One article with SEO title, meta description, excerpt, key takeaways, the stored SEO analysis, publications (where it is live) and optionally the full HTML.
Inputs: article_id (required), project_id, include_html, include_full_analysis.
The Seonix skill for connecting a custom-built site or blog as a publishing channel (endpoints, payload, read routes, media, self-test, hand-over). Same text as https://seonix.ai/SKILL.md. Pass section to get one step.
Inputs: section.
Calendar slots for a month: planned / writing / completed / failed, kind create or optimize, source manual or autopilot, linked article ids.
Inputs: month, project_id.
One cluster with its member keywords, the site pages already covering it, the live SERP snapshot status and the competitor insights Seonix extracted (must-mention entities, recommended blocks, topical coverage, gaps).
Inputs: cluster_id (required), project_id.
An ordered, implementable plan for every open problem: what to change, where (page / site-wide / theme), on which URLs, with the target example. Returned as structured items; pass include_markdown=true to also get a Markdown checklist in the [HIGH] Title — Location / Issue / Fix format. Pages with limit/offset. Items with wp_plugin_fixable / fix_tier auto or ai_assist can be applied by Seonix itself through apply_fix (WordPress). After applying fixes, call start_scan (with the user's yes) and get_site_health again to confirm what cleared.
Inputs: codes, limit, offset, verbose, +5 more.
Status and outcome of an apply_fix run: counts (applied / already correct / no-op / errors) and the per-item results with the before and after values.
Inputs: run_id (required), project_id.
One issue with the evidence the scanner captured on the page (the offending tag, heading outline, redirect chain, JSON-LD block…) and the remediation profile: why it matters, bad vs good example, how-to-fix steps, whether Seonix can fix it through the WordPress plugin.
Inputs: issue_id (required), project_id.
Plan and entitlements for one project: whether the MCP is available, AI generation and publishing permissions, and article quota used/limit.
Inputs: project_id.
Read one Seonix playbook by name (same text as the prompt of that name) — for clients that do not surface MCP prompts.
Inputs: name (required).
One project: site URL, language, autopilot mode and cadence, Search Console connection, publishing channels, plan and quota, SEO progress level, dashboard links. Paid plans only.
Inputs: project_id.
The Seonix remediation profile for an issue code (no project needed): why it matters, bad/good example, fix steps, fix tier, warnings. Use when you meet an unfamiliar code.
Inputs: code (required).
Proof-of-ranking ledger: for each published article, Search Console clicks / impressions / position now vs the previous period, with trend. Requires Search Console.
Inputs: limit, offset, project_id.
Progress of one scan: stage, pages found / scanned, counts, whether background probes are still pending. Poll every 5–10 seconds while status is not completed.
Inputs: scan_id (required), project_id.
The project's SEO Progress ladder: current level and stage, what the next level requires (published articles, quality, Search Console, cluster coverage) and the publishing cadence Seonix recommends for this level.
Inputs: project_id.
The project's brand profile Seonix extracted and the owner refined: what the business does, audience, tone, language, writing instructions, topics to avoid — plus the ready-to-use prompt block. Read it before writing or rewriting anything for the site.
Inputs: project_id.
Site Health for a project: overall score, the four pillars (SEO, AI search, speed, technical), counts of active / fixed / came-back problems, the top open problems by priority, the last scan and the next scheduled check. Start here for any "what's wrong with my site" question.
Inputs: top_tasks, project_id.
Indexing coverage Seonix collected from Search Console: indexed / not indexed / unknown counts, sitemap status, and the pages that are not indexed with Google's verdict.
Inputs: project_id.
Seonix's own impact estimate: current vs potential traffic, keyword coverage, article impacts and top keyword opportunities.
Inputs: project_id.
Keywords already ranking in a position band (default 4–20) with impressions, ordered by the traffic Seonix estimates they would bring at the target position. The classic striking-distance list: refresh those pages first.
Inputs: days, limit, project_id, max_position, +1 more.
The queries one page ranks for (Search Console page+query report filtered to that URL). Use it before rewriting a page's title, meta description or headings.
Inputs: url (required), days, limit, project_id.
Whether Google Search Console is connected for the project, which property, and the last date with published data. Call before other gsc_* tools.
Inputs: project_id.
Daily clicks, impressions, average position and potential traffic for the last N days (7–90), plus the previous-period average position for comparison.
Inputs: days, limit, project_id.
Top landing pages from Search Console for the window: clicks, impressions, CTR, average position.
Inputs: days, limit, project_id.
Top search queries from Search Console for the window: clicks, impressions, CTR, average position.
Inputs: days, limit, project_id.
Articles in the project (Seonix-generated and imported): status, scores, keyphrase, cluster, publish dates. Paginated.
Inputs: page, limit, offset, status, +2 more.
Connected publishing channels (WordPress, Ghost, Webflow, Shopify, Wix, Custom API) with status. A project needs a connected channel before anything can be published.
Inputs: project_id.
Topic clusters (one target page each) with primary keyword, intent, page type, priority, volume, Search Console impressions and winnability. The content plan is built from these.
Inputs: limit, offset, status, project_id.
Persistent issue threads (one per problem per page) with status open / solved / regressed, filterable by code, URL and status. Paginated. Each item carries the issue id for get_issue.
Inputs: url, code, limit, offset, +2 more.
Keywords Seonix researched for the project (seed AI, planner, Search Console, autocomplete, user-added) with volume, competition, Search Console numbers and cluster membership. Paginated; filter by status and substring.
Inputs: limit, query, offset, status, +1 more.
The Seonix playbooks (step-by-step procedures that tell you which tools to call): promote a site, work through a page, fix issues in code, keyword map, internal links, write an article, striking distance, monthly report, connect a blog. Use get_playbook to read one.
Projects (websites) in the account with id, slug, name and site URL. Use the id or slug as project_id in other tools.
Recent site scans with status, pages scanned and severity counts.
Inputs: limit, project_id.
Reverse the changes an apply_fix run made on the live site, restoring the values the plugin snapshotted before applying (write scope — ask the user first).
Inputs: run_id (required), project_id.
Re-run the SEO + readability scoring on an article already in Seonix (no changes saved) and list its problems.
Inputs: article_id (required), project_id.
Start a fresh Site Health scan of the project's site (write scope — ask the user first). Returns the scan id; poll get_scan_status every 10 seconds until status is completed, then call get_site_health / get_fix_plan to see what cleared. One scan at a time per project; the plan's page budget applies.
Inputs: scope, project_id.
Remove the intentional mark from an issue thread so it is scored and counted again (write scope).
Inputs: issue_id (required), reason, project_id.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
AI SEO agent on your Google Search Console data: audits, quick wins, content, WordPress publishing.
AI visibility, Google rankings, Search Console and site audit data for your SEO agent.
Full-cycle SEO automation for AI agents: technical audits, SEO articles, machine-readable pricing.
Site scores, audits, Search Console, keywords, competitors, authority and AI visibility tools.
Add one secure layer between your agents and this server.
