Seonix SEO MCP
OfficialServer Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| PAGESPEED_API_KEY | No | A Google PageSpeed Insights API key. Optional; without it speed audit runs on HTML heuristics only. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| audit_siteA | READ-ONLY, platform-agnostic site auditor. Works on ANY website (WordPress, Shopify, custom, static — anything). Audits SEO, GEO/AEO and speed: it SHOWS problems and gives CMS-neutral 'how it should be' recommendations — it does NOT modify the site; you decide whether/how to fix. Discovers pages via sitemap.xml (polite ~1 req/sec, default 25 / max 100 pages), fetches each page's HTML (following redirects so it can flag redirect chains), and runs ~75 checks: PER PAGE — title missing/length/HTML-entities/lowercase-start, meta description missing/length, image alt text, H1 count, emoji/overlong/before-H1/broken-hierarchy headings, low word count, no internal links, noindex, large page, soft-404, mixed content, canonical, viewport, Open Graph, structured-data validity; GEO/AEO — JSON-LD Article author-not-Person / missing dates / duplicate-type / conflicting-data / unrecognized @type / incomplete Person, FAQ & HowTo schema not visible on-page, incomplete social tags, AI-restrictive meta robots, /llms.txt, robots.txt AI-bot blocking, sitemap lastmod coverage; CROSS-PAGE / CRAWL — duplicate titles & meta descriptions, trailing-slash duplicates, boilerplate repeated headings, broken internal links, orphaned pages, crawl depth, redirect chains (broken/loop/too-many); SITE — robots.txt missing/blocks-all/blocks-sitemap-path, sitemap unreachable/invalid/empty/index-children-failed/not-declared-in-robots; SPEED — always-on HTML heuristics (render-blocking head resources, images missing width/height, un-lazy offscreen images, large inline blocks, page weight, DOM size), plus Google PageSpeed Insights (Core Web Vitals + Lighthouse opportunities) on a sample of pages when PAGESPEED_API_KEY is set. Returns a per-pillar summary {seo, aeo, speed} (score + health_label + issue count) and a flat issues[] where each item = {code, category, severity, url, evidence, why, target_state, recommendation}. |
| speed_auditA | READ-ONLY speed-only audit of ONE page. Always runs the HTML speed heuristics on the fetched page; when PAGESPEED_API_KEY is set it also runs Google PageSpeed Insights (mobile) for that page and returns Core Web Vitals (LCP, CLS, INP, TBT, TTFB) plus the top Lighthouse opportunities. Every finding carries category 'speed' and a 'how it should be' recommendation. Use audit_site for a whole-site, multi-pillar audit. |
| propose_fixesA | READ-ONLY safe-fix ADVISOR for ONE page. Audits the page, then proposes a concrete, minimal fix for each issue: what to change, whether the change is VISIBLE or INVISIBLE to visitors, which issue code(s) it clears, and safety notes. Deterministic fixes carry an exact edit (inject viewport / Open Graph tags, decode HTML entities in the title, strip a leading heading emoji, rewrite http:// → https://, retag a heading that skips a level). Others are classified: 'needs-value' (alt text, dates, og:image — mechanical but a value must be supplied), 'manual' (a content/structure decision, guidance only), or 'infra' (server / robots.txt / sitemap / CDN / speed — not page markup). It NEVER writes anything. Pass any returned proposal to preview_fix to dry-run it before deciding. |
| preview_fixA | READ-ONLY dry-run of a single fix proposal (the object returned by propose_fixes) against the page's CURRENT HTML. Applies the edit statically — no rendering, no writing — and returns a regression-gate verdict: 'pass' (the edit is localized and touches only its intended region), 'idempotent' (already applied — a no-op), 'blocked' (the target is ambiguous or missing, so an automatic edit is unsafe), or 'manual' (needs a value, or lives in infrastructure). Includes the before/after of the affected region. This is a STRUCTURAL safety gate (does the edit stay in its region?), not a pixel/visual gate. NEVER modifies the site. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool has a clearly distinct role: site-wide audit, per-page fix proposal, dry-run preview, and speed-only audit. No overlapping purposes; an agent can easily select the appropriate tool.
All tool names follow a consistent verb_noun pattern in snake_case (audit_site, preview_fix, propose_fixes, speed_audit). No mixing of conventions or ambiguous verbs.
Four tools is appropriate for an SEO analysis MCP server: site audit, fix proposal, fix preview, and speed audit. The scope is focused and each tool earns its place.
The tool set covers the full analysis workflow: comprehensive audit, targeted fix proposals, safety preview, and speed-specific audit. No obvious gaps given the read-only advisory nature.