Jsonify SEO Refresh

Run SEO refresh batches on Jsonify use case pages. This skill is for repo-native execution: inspect the current repo state, use Google Search Console when Mode A needs ranking data, skip any page touched in the last 90 days according to seo-page-metrics-history.csv, edit the allowed fields only, record a metrics snapshot for every touched page, update the batch log, and finish with one PR per batch.

When to use

• The user asks for a copy refresh batch, SEO refresh batch, or Mode A batch.
• The user asks for an indexability unlock batch, noindex cleanup, or Mode B batch.
• The user wants help selecting /usecases/<slug> candidates or updating use case metadata for SEO.

Hard boundaries

• Only edit these five frontmatter fields in _usecases/<slug>.md: seo_title, meta_description, title, description, sources_text.
• Only edit indexability in _data/usecase-seo.json.
• Do not touch _includes/meta.html, layout files, or templates.
• Do not create new use case markdown files.
• Do not add new keys to existing SEO JSON entries.
• Do not run npm run seo:compile-usecase-seo after making batch edits. This workflow treats _data/usecase-seo.json as the working control file for the batch.

Quick workflow

1. Read references/operating-spec.md before the batch if you need the exact selection rules, copy rules, or PR format.
2. Determine the mode.
3. Read seo-page-metrics-history.csv and skip any slug touched in the last 90 days. Use seo-batch-log.md as audit history, not as the cooldown source.
4. For Mode A, shortlist candidates with scripts/mode_a_candidate_queue.mjs.
5. For Mode B, shortlist review pages with scripts/mode_b_review_queue.mjs.
6. Read the target _usecases/<slug>.md files and the relevant entries under _data/usecase-seo.json.pages.
7. Write the five copy fields, and apply JSON indexability changes only when the mode requires them.
8. Append a dated metrics snapshot for every touched slug to seo-page-metrics-history.csv.
9. Update seo-batch-log.md.
10. Open one PR for the batch.

Commands

Mode A queue

node codex-skills/jsonify-seo-refresh/scripts/mode_a_candidate_queue.mjs --days 90 --limit 8

Mode B queue

node codex-skills/jsonify-seo-refresh/scripts/mode_b_review_queue.mjs --limit 8

Metrics history snapshot

node codex-skills/jsonify-seo-refresh/scripts/append_metrics_history.mjs --spec-file /tmp/seo-batch-spec.json --from-gsc /tmp/jsonify-usecases-gsc.json

Operational notes

• query_gsc_usecases.mjs normalizes trailing-slash variants and emits both raw page-query rows and page summaries.
• mode_a_candidate_queue.mjs applies the Mode A ranking heuristics and skips slugs touched inside the 90-day cooldown window.
• mode_b_review_queue.mjs derives verticals from slug keywords, skips recently touched pages, and flags likely overlap with existing canonicals so you can avoid promoting near-duplicates.
• append_metrics_history.mjs appends one CSV row per touched slug to seo-page-metrics-history.csv, including snapshot date, GSC date window, action, mode, primary keyword, page state, robots, and page-level GSC metrics.
• In _data/usecase-seo.json, batch work should only touch entries inside pages. Leave top-level metadata such as generatedAt, summary, sources, and warnings alone.
• Prefer direct, outcome-led copy for pricing analysts and CI leads. The full copy guardrails live in references/operating-spec.md.

Reference map

• references/operating-spec.md: full workflow, copy rules, vertical rules, JSON update rules, and PR format.
• scripts/query_gsc_usecases.mjs: GSC auth + pull for /usecases/ pages.
• scripts/mode_a_candidate_queue.mjs: Mode A queueing with GSC data plus 90-day cooldown filtering.
• scripts/mode_b_review_queue.mjs: deterministic Mode B candidate queue from repo data.
• scripts/append_metrics_history.mjs: append metrics history rows for updated or promoted pages.