mcpField
S
seo-ai-social-report

seo-ai-social-report

github.com/serankingMarketing & growth18958 installsUpdated 18w ago
Repository

About seo-ai-social-report

A reporting skill that combines search rankings and AI visibility from SE Ranking with social engagement from Planable. It produces concise written conclusions and an interactive HTML report for teams that need a shared view of these channels.

Key features

  1. 01SEO ranking and traffic reporting
  2. 02AI-search visibility trends
  3. 03Planable social metrics
  4. 04Interactive charts and KPI cards
  5. 05Text conclusions and top-post highlights

Use cases

  1. 01Create a monthly cross-channel report
  2. 02Review search and AI visibility trends
  3. 03Compare social engagement and top posts

MCP support

Requires SE Ranking MCP for SEO and AI-search data and Planable MCP for social metrics and posts. The skill gathers data from both connected services for the report.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/seranking/seo-skills --skill seo-ai-social-report

After installing, type/seo-ai-social-reportin Claude Code to confirm it loaded.

Details

From the project README

Unified SEO + AI + social report

Most reporting forces clients to read three tools separately. This produces one view: search rankings and AI-search visibility from SE Ranking, social engagement from Planable, with quick text conclusions for the inbox and an interactive HTML report for the detail.

Prerequisites

  • SE Ranking MCP connected (a project gives the cleanest ranking trend; Data API history works without one).
  • Planable MCP connected, with the workspace and connected pages.
  • The user provides: domain + brand, market/country (default us), the Planable workspace, the reporting period (default: previous calendar month), and optionally competitors for share-of-voice.

Connector health check

Before doing anything else, verify both MCPs are reachable:

Only continue to the process steps below once both calls return a successful response.

Process

1. Scope

Confirm domain, Planable workspace, period, and competitors. Resolve the period to explicit start/end dates.

2. Pull SEO (SE Ranking)
  • With a project: PROJECT_getSummary and PROJECT_getPositionHistory(site_id, type: avg_pos | visibility, date_from, date_to) for the ranking trend; optionally reuse a finished audit (PROJECT_listAudits → getAuditReport) for a site-health number.
  • Without a project: DATA_getDomainOverviewWorldwide for the snapshot.
  • Shortcut: the AI overview call in step 3 (time-series view) already returns monthly organic_traffic / overall_traffic streams for the chosen country — you can reuse those for the organic-trend chart instead of a separate history call.
3. Pull AI search (SE Ranking)
  • DATA_getAiSearchOverview(target, source, brand?) with the default time-series view — brand presence, link presence, AI opportunity traffic, average position, plus monthly trend streams (including organic/overall traffic). If previous is null, treat it as a baseline — don't render the change_percent of 100 as growth.
  • DATA_getAiSearchLeaderboard(primary, competitors[], source, engines[]) — share of voice vs competitors (only if competitors were provided). Query it narrowly (few engines/competitors per call) — it can time out on large requests.
4. Pull social (Planable)
  • list_pages(workspaceId) → page IDs.
  • get_page_metrics_summary(workspaceId, pageIds, startDate, endDate) — account-level audience/impressions/engagement-rate.
  • get_post_metrics_summary(workspaceId, pageIds, startDate, endDate) — published-post totals in the window. (Different scope from page impressions — keep them separate.)
  • get_post_metrics(workspaceId, pageIds, ..., limit: 5) — top posts.
  • list_posts(...) — posts published in the period (volume/cadence).
  • For a monthly social trend, call get_page_metrics_summary once per month bucket.
5. Write the quick conclusions (chat)

A ~10–15 line plain-text summary: the headline move in each of the three areas (SEO, AI search, social), then 2–3 specific callouts ("what moved", "what to watch"). This is the part most people actually read — make it sharp and number-led, no filler.

6. Build the interactive HTML report

Use the bundled template at assets/report-template.html as the starting point — it has the Planable colour palette, the card/section layout, and Chart.js (CDN) wired up, including a dual-axis trend chart. Populate it with the real data you pulled:

  • KPI cards: organic traffic, AI presence / share of voice, social impressions, engagement rate, posts — each with a real period-over-period change where you have one (leave the delta blank otherwise; never invent it).
  • Trend charts (interactive): ranking/organic trend, AI presence trend, social impressions & engagement. When two series have very different scales (e.g. organic traffic in the thousands vs AI opportunity traffic in the hundreds, or impressions vs engagement), plot them on a dual axis or in separate charts — don't squash them onto one axis or the smaller line flatlines.
  • Share of voice: a bar chart of brand vs competitors (omit if no competitors).
  • Top posts and What moved / What to watch — same callouts as the text summary.

Inline the data as JavaScript arrays in the file (no external data files, no localStorage). Save to the outputs folder as {client}-{period}-cross-channel-report.html, then share it with present_files. Validate the inline script parses (a quick node --check-style pass) before presenting.

Output

  1. Quick conclusions — short text summary in chat (the numbers + 2–3 callouts).
  2. Interactive HTML report — self-contained file, presented via present_files, with KPI cards, hoverable trend charts, share-of-voice, and top posts.

Tips

  • Keep the chat summary tight; the HTML is the detailed artefact. Don't duplicate the full report in chat.
  • Label data sources in the report footer (SE Ranking for search/AI, Planable for social) and stamp the generation date and exact date range.
  • Round for readability (48K, 4.2%); never fabricate a trend. If a series has one data point, show the point and skip the trend line.
  • Mark partial periods (e.g. a month-to-date bucket) clearly so a short bar isn't read as a drop.

Edge cases & limits

  • No GA4 / no Looker Studio. This report covers what the two MCPs return — search/AI visibility and social engagement, not cross-channel conversion attribution. State this in the footer.
  • Limited page-level social analytics: Twitter/X, Pinterest, Threads and Google Business Profile typically don't return page-level metrics via the connector — they appear in unsupportedPages; base totals on supported channels and note the rest as "data not available".
  • Thin periods: if little was published or rankings barely moved, say so plainly rather than dramatising noise.
  • No competitors provided: skip the share-of-voice section instead of leaving an empty chart.

Comments

Markdown supported