Programmatic SEO (quality-first)
Talk like ../../references/plain-language.md. Ideas adapted from MIT pSEO skill sets (see ../../references/ATTRIBUTIONS.md).
Positioning
pSEO is useful when each page answers a distinct search intent with enough unique data. It is not a way to flood the index with near-duplicates.
Hard vetoes (refuse and explain)
- Doorway / spun / synonym-swapped pages
- Pages with no unique facts beyond the city or product name
- Unreviewed scaled AI filler (Google spam-update risk)
- Rented subfolders on unrelated high-DR domains ("parasite" spam)
Web connector mode
Design the data architecture and quality gates in chat. Sample drafts: check_ai_content. Live SERP/keyword checks: quote_cost → yes → call-actor on khadinakbar/*. Citebase does not auto-publish page farms.
Workflow
- Discover — Which entities × attributes have enough first-party or licensed data? (cities × services, integrations × features, comparisons with real specs.)
- Data architecture — One row per page; required fields; sufficiency gate (skip rows missing required facts).
- Templates — Unique H1/title pattern, answer-first intro, sections driven by data columns, not filler paragraphs.
- Quality gates before publish:
- Intent uniqueness (no cannibalization with an existing money page)
- Minimum unique tokens / facts per page
ai-content-check/check_ai_contenton sample draftseeat-auditsample on YMYL niches
- Rollout — Gradual indexing (sitemap batches), internal hub links, monitor thin-content flags via
seo-audit. - Live SERP/keyword checks only after
quote_costonkhadinakbar/*. - Done / Cost so far / Next.
Hand-offs
- Entity graph for the niche →
entity-seo - Single-page rewrite →
content-optimize/humanize-content - Briefs for hub pages →
content-brief