agentsclimarketplace

Seo performance tracker

Skill siddiqss/semantic-seo-suite/skills/seo-performance-tracker

Grounded semantic SEO, GEO & off-page as Claude Code skills — with a fabrication guard that refuses to invent numbers. Free & MIT.

Install
npx -y skills add siddiqss/semantic-seo-suite --skill seo-performance-tracker

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Track topical-map performance from Google Search Console — per-pillar impressions/ clicks/position rollups, striking-distance opportunities, decaying pages, measured cannibalization, and GSC queries the map doesn't yet cover. Use whenever the user asks how content is performing, wants a GSC/Search Console report, mentions rankings, impressions, clicks, or asks what to update or write next based on real data. Requires GSC access; degrades honestly if it's not configured. Triggers on performance/analytics intent broadly.

SKILL.md

3.4 KB, as published. Nobody here has run it

seo-performance-tracker

The measured feedback loop. It replaces guesswork (and the fabricated "Pillar Page Rank" that wrapper tools invent) with real Search Console data joined to the topical map. Every figure is measured (GSC) or derived (rollups) — never estimated.

Read first: ../../framework/topical-map-theory.md (pillars/priority), ../../framework/query-semantics.md (query networks → new nodes).

Preconditions

  • topical-map.json with urls on published nodes (the auditor sets these).
  • GSC configured: grounding.sources.gsc: true + credentials.gsc_credentials_path. If not configured, say so and do NOT invent metrics — spot-checking positions via web_search is unreliable; be explicit about that limitation.

Workflow

  1. Pull GSC data:

    python ../../scripts/gsc_client.py --site "<property>" --creds <path> \
  --start <date> --end <date> --dims query,page \
  --out brands/<slug>/data/gsc/pull.json
# optional prior period for decay:
python ../../scripts/gsc_client.py ... --start <earlier> --end <earlier> \
  --out brands/<slug>/data/gsc/prev.json

  2. Analyse against the map:

    python ../../scripts/gsc_analyze.py --gsc brands/<slug>/data/gsc/pull.json \
  [--gsc-prev brands/<slug>/data/gsc/prev.json] \
  --map brands/<slug>/topical-map.json --brand "<Brand>" \
  --out brands/<slug>/audits/<date>-performance.md

    Produces:

    • Pillar rollups — impressions/clicks/avg-position per pillar (derived). The honest Pillar Page Rank.
    • Striking-distance — queries at position 5–15 → quick-win update targets.
    • Measured cannibalization — one query landing on multiple pages (the strongest cannibalization signal; confirms/ː refutes the auditor's embedding guess).
    • Decaying pages — >30% click loss vs the prior period.
    • Uncovered queries — GSC queries not in any node's query network → candidate new map nodes.

  3. Feed the loop:

    • Add uncovered-query candidates to the map via topical-map-builder (as new nodes or as query-network additions to existing nodes).
    • Mark striking-distance and decaying nodes needs-update; push them up calendar.md.
    • Send confirmed measured-cannibalization pairs to semantic-site-auditor / linking-and-schema for consolidation/redirect.

Definition of done

  • Performance report produced with all five sections, every figure provenance-tagged.
  • Actionable next steps flowed back into map + calendar.
  • If GSC isn't configured: an honest "not available" with the setup pointer, no invented numbers.

Grounding ladder

  • T1 (GSC): the whole skill — this is a T1 capability (GSC is free).
  • Without GSC: the skill explains it can't measure performance reliably and points to README GSC setup. It does not fabricate a substitute.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.