Plan Programmatic SEO Pages Without Thin Content

Unlike free "write me 500 programmatic pages" prompts and page generators, this skill runs a quality gate on your dataset before you build anything. A bundled script renders the URL, title, H1 and meta for every row, then marks each page PUBLISH, FIX or HOLD for duplicate URLs, thin data, near-duplicates, length and zero demand. Every HOLD comes with a reason you can fix in the data. It never pads pages with filler.

Sample run: a "best MCP servers for SEO" directory, one page per server, built from the official MCP Registry (registry.modelcontextprotocol.io, search "seo", 234 latest active servers, fetched 21 Oct 2026; public data; no demand column, so demand is left blank and never estimated):

1. Run 1, URL /mcp-servers/seo/{name}: HOLD 67, including 58 duplicate URLs (21 servers are named "site", 13 "seo", 8 "seo-audit"), so the URL was changed to publisher + name
2. Run 2: 234 rows | PUBLISH 188 | FIX 32 | HOLD 14
3. HOLD near-duplicate: com.seoapi/googlebot, bingbot, yandexbot and 4 more (one-line "Validate X IP addresses" servers, similarity 0.85-0.90)
4. HOLD near-duplicate: uk.co.seoplumbers/site, uk.co.seodealerships/site, 2 more (same "the site's own MCP server" text, 0.81-0.82)
5. HOLD thin: io.github.sparrow84001/mcp-seo (only the description counts: no install info, and its one repo URL is under the 2-word minimum)
6. FIX 32 titles over 60 chars (e.g. "SEO Analytics & Insights — Google Search Console MCP MCP Server for SEO: Tools + Setup", 86)
7. Run 3 with max_similarity 0.70: HOLD 33, now 14 of the 18 seoapi validators and 14 of the 21 UK "site" clones
8. Passed the gate but off-topic: paseo, Seoul apartment signals, a semver checker. The registry search "seo" matches substrings. The gate can't judge relevance, so that is the human review step
9. Not caught by the script: 21 pages share the title "site MCP Server for SEO: Tools + Setup", and 9 PUBLISH titles say "MCP MCP Server" (e.g. "DataForSEO MCP MCP Server for SEO…"). Fix the title template, then re-run

Full report from this run in the Example below.

Works in Claude Code, Cursor and Codex · tested on real public data (the MCP Registry run above) · SKILL.md + one Python script (stdlib only, no network, no API keys) + templates · never invents search demand or traffic forecasts · v1.0, updated Oct 2026 · questions answered via Agensi messages within 24h · Check the search intent behind your page pattern with our Apify actor: https://apify.com/cygnix_labs/serp-intent-mapper · Free no-API-key SEO MCP server (keyword clustering, search intent, briefs): https://github.com/cygnixlabs-cmd/cygnix-seo-mcp

Programmatic SEO can bring in hundreds of pages' worth of traffic, or get a whole section ignored as scaled content abuse. The difference is whether each page has real data of its own, and most teams only find out after launch.

What it does

• Renders URL, title, H1 and meta for every row from a simple JSON template spec ({column} and {column_slug} placeholders).
• Holds duplicate URLs, thin rows (too few unique fields with real data), near-duplicates (word overlap of the unique data above your threshold) and rows with zero demand.
• Flags titles and metas over length, missing placeholder data and missing demand data as FIX.
• Renders sample page outlines so a person can check the pages are actually useful.
• Plans a phased launch: 10-20% of the PUBLISH rows first, a hub page, and a 4-6 week stop rule in Search Console.

How it works

1. Pick the page pattern and data source (templates/pseo-opportunity-brief.md).
2. Build the dataset CSV (one row per page) and the template spec (templates/page-template-spec.json).
3. Your agent runs scripts/pseo_planner.py, then reviews the HOLD and FIX rows and the sample outlines with you (templates/quality-gate-checklist.md).
4. You get pages-plan.csv (verdict, URL, title, H1, meta, lengths, unique data filled, closest duplicate, demand, reasons), sample-pages.md and a launch plan.

Who it's for

SEO leads, SaaS growth teams and directory or marketplace builders planning integration, location, comparison or "X for Y" pages from a spreadsheet. It is not a page generator: it decides which rows deserve a page.

Pairs well with (Cygnix Labs, on Agensi)

• Keyword Clusters + Intent Map: No Paid SEO Tool Needed (agensi.io/skills/keyword-clusters-intent-map-no-paid-seo-tool-needed): find the page pattern
• SERP Intent Mapper — Keyword → Page Type Match (agensi.io/skills/serp-intent-mapper-keyword-page-type-match): confirm the page type for the pattern
• Local SEO Keyword Pack — Service × City Matrix + Page Plan (agensi.io/skills/local-seo-keyword-pack-service-city-matrix-page-plan): build a service × city dataset
• Map Niche into Pillars Clusters and Internal Links (agensi.io/skills/map-niche-into-pillars-clusters-and-internal-links): plan the hub page
• SEO Meta Tags Generator — Titles, Descriptions + Schema (agensi.io/skills/seo-meta-tags-generator-titles-descriptions-schema): rewrite the title template the gate flags
• Find Orphan Pages and Fix Your Internal Linking (agensi.io/skills/find-orphan-pages-and-fix-your-internal-linking): link the new pages after launch