A complete set of skills for implementing programmatic SEO (pSEO) at scale. Works with Claude Code, Cursor, GitHub Copilot, Windsurf, and any LLM-powered coding assistant. Designed for 1000+ page sites with built-in quality safeguards, memory management, and compliance with Google's 2025 algorithm updates and LLM visibility optimization.
| Skill | Purpose |
|---|---|
| pseo-discovery | Analyze codebase and business context to find pSEO opportunities |
| pseo-audit | Audit codebase readiness for pSEO at scale (read-only) |
| pseo-data | Design the structured data architecture powering all pages |
| pseo-templates | Build page templates with unique, intent-matched content |
| pseo-metadata | Implement dynamic title, description, canonical, OG, and Twitter tags |
| pseo-schema | Add JSON-LD structured data (Article, FAQ, Breadcrumb, Product, etc.) |
| pseo-linking | Build internal linking: hub-spoke, breadcrumbs, related pages, sitemap, redirects |
| pseo-performance | Optimize builds, Core Web Vitals, memory, and publication velocity |
| pseo-llm-visibility | Optimize for AI citation: llms.txt, AI crawlers, content chunking, entity optimization |
| pseo-quality-guard | Validate against thin content, duplicates, cannibalization, and scaled content abuse |
| pseo-scale | Architect for 10K-100K+ pages: database layer, data sufficiency gating, incremental validation, crawl budget, CDN |
| pseo-orchestrate | Coordinate all skills in the correct dependency order |
pseo-discovery (0)
|
v
pseo-audit (1)
|
v
pseo-scale (1.5) <-- conditional: only if 10K+ pages
|
v
pseo-data (2)
|
+---------------+
v v
pseo-templates (3) pseo-linking (4)
| |
+-------+-------+
|
+-------+-------+
v v
pseo-metadata (5) pseo-schema (6)
| |
+-------+-------+
|
+-------+-------+
v v
pseo-performance pseo-llm-visibility
(7) (8)
| |
+-------+-------+
v
pseo-quality-guard (9)
Each skill is a self-contained markdown file in the skills/ directory. Copy to your AI assistant's custom instructions location:
| Assistant | Copy to |
|---|---|
| Claude Code | .claude/skills/ |
| Cursor | .cursor/rules/ (rename to .mdc) or .cursorrules |
| GitHub Copilot | .github/copilot-instructions.md (concatenate) |
| Windsurf | .windsurf/rules/ or .windsurfrules |
| Other | Feed as system prompts or custom instructions |
The skills are plain markdown and work with any assistant that accepts custom instructions.
Ask your AI assistant to run individual skills or the full pipeline:
- Full pipeline: "Run the pseo-orchestrate skill with full pipeline"
- Discovery only: "Run pseo-discovery to find pSEO opportunities"
- Single skill: "Run pseo-data to design the data architecture" or "Run pseo-quality-guard to check for abuse patterns"
The orchestrate skill handles dependency ordering automatically. For manual execution, follow the pipeline order shown above.
- Scales to 100K+ pages: Database-backed data layer, data sufficiency gating, incremental validation, CDN/edge delivery, crawl budget management
- Two-tier data architecture for memory safety (PageIndex vs BaseSEOContent), cursor-based iteration at scale
- Data sufficiency gating: Prevents thin page generation before build time — only pages with enough data get created
- Google 2025 compliance: E-E-A-T signals, Needs Met enforcement, scaled content abuse detection, YMYL risk assessment
- LLM visibility: llms.txt, AI crawler configuration, answer capsule pattern (134-167 words), entity optimization
- Scalable similarity detection: MinHash/LSH and SimHash instead of O(n^2) pairwise comparison
- Incremental validation: Delta-only quality checks for fast iteration at scale, with periodic full scans
- Publication velocity controls: Gradual rollout strategies from 100 to 100K pages
- Content enrichment pipeline: Structured data enrichment with optional LLM-assisted generation and human review sampling
- Framework-agnostic: Next.js App Router by default, adaptable to Astro, Nuxt, Remix, SvelteKit
Created by mariluukkainen