Plans a programmatic SEO project: demand, the data each page needs, the template, quality rules, URL structure and a rollout. And it tells you when the idea is thin content.
Programmatic SEO works when every page has something real on it, and gets you a manual action when it doesn’t. This skill plans the version that works, or tells you plainly not to bother.
What it does
Checks demand for the page pattern and its variations
Lists the unique data each page needs, and stops you if there isn’t enough
Lays out the template, the quality bar and the internal linking
Plans a small first batch before scaling, including multilingual pages
Try asking
“Plan programmatic pages for “[our app] + [integration]”. We have 300 integrations.”
“Is “email marketing for [industry]” a good pSEO pattern for us, or thin content?”
Install it
Download the ZIP above.
Claude.ai: open Settings, go to Capabilities and upload it under Skills.
Claude Code: unzip it into ~/.claude/skills/ (all projects) or .claude/skills/ (one project).
The full instructions are below. It’s one Markdown file, so change anything you like.
Inside the skill
programmatic-seo-planner/SKILL.md
---
name: programmatic-seo-planner
description: Plan a programmatic SEO project (many pages from one template and a dataset), such as location, integration, comparison, template or multilingual landing pages. Use when someone wants to scale pages programmatically, asks about pSEO, or wants to know if a page type can be templated without becoming thin content.
---
# Programmatic SEO planner
Turn an idea for "hundreds of pages" into a plan that's worth indexing, or say clearly that it isn't.
## Inputs
- The **page pattern** (e.g. "<tool> integration with <app>", "<service> in <city>", "<product> vs <competitor>")
- What the business sells and who searches for it
- The **data** they have or can get
- Keyword data if available (volumes per variation); otherwise estimate and say so
## Steps
1. **Demand.** Check that people search the pattern, with real variations. Group the long tail into the head term, modifiers and the number of combinations.
2. **Data.** For each page, what unique information fills it? List the fields. If most pages would differ only by a swapped word, stop and say so. That's thin content.
3. **Template.** Lay out the page: the H1 pattern, the above-the-fold answer, the data-driven sections, FAQs, CTAs and internal links. Mark which parts are unique per page and which are shared.
4. **Quality bar.** Set rules for when a page is published: minimum data fields, minimum unique text, and a human review sample. Pages below the bar stay unpublished or noindexed.
5. **Structure.** URL pattern, hub pages, breadcrumbs, internal links between siblings and back to the hub, and sitemap handling.
6. **Rollout.** Start with 20–50 pages, measure indexing and traffic for 4–6 weeks, then expand the patterns that work.
7. **Languages** (if multilingual): translate with review, set hreflang, and localise the data and examples, not just the words.
## Output format
```
## Verdict
<go / go with changes / don't, in two sentences>
## Pattern and demand
## Data fields (unique per page)
## Page template
## Quality rules
## URL and linking structure
## Rollout plan
## Risks
```
## Rules
- Be blunt about thin-content risk. Google's spam policies call out scaled content made mainly to rank.
- Never suggest publishing thousands of pages at once.
- Mark estimates and assumptions clearly.
Questions
How do I install it in Claude.ai?
Download the ZIP, then in Claude.ai open Settings, go to Capabilities and upload it under Skills. Claude picks it up when your request matches what the skill does.
How do I use it in Claude Code?
Unzip it into ~/.claude/skills/ for every project, or into .claude/skills/ inside one project. The folder name stays the same as the skill's name.
Can I edit it?
Yes, it's a plain Markdown file (SKILL.md). Change the steps, the output format or the rules to match how your team works.