SEO Content Briefs skill for Claude Code

Briefs built from the SERP, not from keyword density. Claude reads the real top results before planning a page, records format consensus and SERP features like AI Overviews, and checks your site for cannibalization. The brief it hands over names one information-gain angle, an E-E-A-T evidence plan, a question-led outline, internal links in both directions, and metadata, so writers never need a keyword-count target.

  • nameseo-content-briefs
  • size196 lines · ≈2,200 tokens
  • files1
  • licenseMIT
  • updated

What Claude sees before loading it

This description is the only part of the skill that sits in context all the time. Claude reads it to decide when to load the full playbook below.

description: Produces SEO content briefs from live SERP analysis, covering search intent, dominant content format, the angle that adds information gain, outline with H2/H3s, entities and questions to cover, E-E-A-T evidence to include, internal links, and on-page metadata. Use when planning a blog post, guide, comparison, or landing page meant to rank, when a writer needs a brief, when refreshing a page that lost rankings, or when asked what to write to target a keyword.

Try asking

  • Write a content brief for 'postgres vs mysql for startups' that can beat the current top 5
  • Our guide to invoice factoring dropped from #3 to #14. Brief a refresh
  • I need briefs for our freelance writers for these 5 keywords. Start with the first one

Use it when

  • Planning a new article, guide, or comparison page meant to rank
  • Handing work to a writer who needs research done up front
  • Refreshing a page that has lost rankings or clicks
  • Deciding whether a keyword needs a blog post or a product page

Skip it when

How to check the result

The brief cites the observed SERP format consensus, names a concrete information-gain angle with its evidence source, and every outline section maps to a searcher question or entity.

The common mistake it prevents

Copying the outline of the current top results, which guarantees a page with nothing new to offer and no reason to outrank them.

~/.claude/skills/seo-content-briefs/SKILL.md

--- name: seo-content-briefs description: Produces SEO content briefs from live SERP analysis, covering search intent, dominant content format, the angle that adds information gain,… license: MIT ---

A good brief lets a writer produce a page that matches what searchers want, says something the current top results do not, and proves first-hand experience — without the writer redoing the research. Length targets and keyword counts are not the point; intent match, information gain, and evidence are.

Ground rules

  • Look at the actual SERP for the target query (in the target country and language) before writing anything. Intent is read from results, not guessed from the keyword.
  • One brief = one primary intent. If the top 10 split between two intents, that is two pages or a deliberate hybrid — decide explicitly.
  • Never instruct “use the keyword N times” or target a keyword density. Ask for natural coverage of the topic, entities, and questions.
  • Every brief names what this page will contribute that the top results lack (original data, tested steps, screenshots, expert quotes, a decision framework, a calculator).
  • Write for people first. Google’s helpful-content signals are part of core ranking; content produced mainly to rank, at scale, without added value is a spam-policy risk.
  • AI-generated drafts are allowed by Google if helpful; the brief must still demand first-hand material an LLM cannot invent. Never fabricate experience, quotes, or stats.

Workflow

- [ ] 1. Confirm the target query, market, and page type
- [ ] 2. Check for cannibalization on the existing site
- [ ] 3. Analyze the SERP (intent, format, features, gaps)
- [ ] 4. Collect questions and entities
- [ ] 5. Define the angle and the evidence plan
- [ ] 6. Write the outline
- [ ] 7. Specify metadata, links, and schema
- [ ] 8. Deliver the brief in the template

1. Confirm inputs

Primary keyword, 3–8 close variants (same intent), country/language, target URL (new or existing), business goal (signups, demo, affiliate, awareness), and the product’s real relationship to the topic. If keywords are not chosen yet, run keyword-research first.

2. Check cannibalization

site:example.com "primary keyword"

Also check Search Console → Performance → Queries filtered by the keyword: if two URLs swap impressions for it, consolidate (merge + 301) or differentiate intents before adding a third page. If an existing page already ranks 5–20, brief a refresh of that URL instead of a new page.

3. Analyze the SERP

Record the top 10 organic results plus SERP features. Use a table:

#URLContent typeFormatAngle / hookFreshnessNotable gaps
1guide / listicle / tool / product / forumhow-to, comparison, definition“for beginners”, “2026”, “tested”date

Then decide:

Signal in SERPRead it asBrief implication
Mostly product/category pagesTransactionalLanding/category page, not a blog post
Mostly guides and how-tosInformationalStep-by-step guide with visuals
“Best X”, “X vs Y” listiclesCommercial investigationComparison with criteria, tested picks, verdict
Reddit/forums prominentSearchers want lived experienceLead with first-hand testing, honest trade-offs
AI Overview presentQuery is answerable in a paragraphPut a crisp 40–60 word answer early; win the click with depth the summary lacks
Video carouselVisual taskEmbed or produce a short video; add VideoObject
Local packLocal intentLocation page, not an article
Featured snippet (list/table/paragraph)Format Google prefersMirror that format under a question-shaped H2
All results < 6 months oldFreshness mattersDated content, update cadence in brief

Format consensus (6+ of 10 results share a format) is a strong constraint; deviate only with a reason. Word count: note the range of the top 5 as context only — never as a target.

4. Collect questions and entities

  • People Also Ask (expand 2–3 levels), related searches, autocomplete variants.
  • Forum threads and community questions for the real objections and edge cases.
  • Entities: the products, people, standards, metrics, and concepts the top results all mention. List them as “must explain” vs. “mention”. Missing core entities signals a shallow page.
  • Sales/support tickets and product docs: questions competitors cannot answer.

Group questions under the outline section where they belong; do not dump a FAQ at the end.

5. Define angle and evidence (information gain)

Pick one primary differentiator and state it in a sentence:

  • Original data or benchmark you ran (“we tested 14 tools on the same 10 GB dataset”).
  • First-hand process with real screenshots, configs, failure cases.
  • Expert input with name and credentials (interview, review, quote).
  • A framework or decision table that resolves the reader’s choice.
  • Interactive asset (calculator, template, checker).

E-E-A-T evidence plan: named author with relevant experience and a bio page, reviewer for YMYL topics (health, finance, legal, safety), sources cited inline to primary documents, dates of testing, disclosure of affiliate or product relationships.

6. Write the outline

  • H1 matches the intent and includes the primary topic naturally.
  • Intro: answer or promise within the first 2–3 sentences; no throat-clearing.
  • H2s follow the reader’s sequence of questions; phrase some as the actual question when a snippet or AI answer is likely.
  • Under each H2: the point to make, the evidence to show, entities/questions to cover, and the visual (table, screenshot, diagram).
  • Comparison pages: criteria table up front, then per-option sections, then verdict by use case.
  • End with the next action that matches the business goal, not a summary.
  • Title tag: ≤ ~60 characters (~580 px), primary topic first, a reason to click (year only if the content is genuinely dated). Google may rewrite titles that do not match the H1 or content.
  • Meta description: 140–160 characters, the promise plus a specific detail; not a ranking factor but affects clicks.
  • URL slug: short, lowercase, hyphenated, no dates unless news.
  • Internal links out: 3–6 contextual links to related pillar/cluster pages with descriptive anchors. Internal links in: 2–4 existing pages that should link to this one (find with site:example.com topic).
  • External links: cite primary sources.
  • Schema: Article/BlogPosting with author; VideoObject if a video is embedded. Hand to schema-markup for implementation.

Refresh briefs

When a page decays (Search Console clicks down > 20% over 3 months vs. prior period, with stable demand), brief a refresh: re-run the SERP analysis, list what newer top results added, update facts/screenshots/dates, merge overlapping thin posts into it, keep the URL. Change dateModified only for substantive updates.

Gotchas

  • Briefing from a keyword tool’s volume alone, without opening the SERP, routinely produces a blog post for a query Google answers with product pages.
  • “Ultimate guide” briefs that cover everything rank for nothing; cut sections that serve a different intent and link to them instead.
  • Copying the top results’ outline produces zero information gain. The outline should cover what they cover and the gap.
  • Do not ask writers to add FAQ sections for FAQ rich results; those are limited to government and health sites.
  • AI Overviews answer simple definitions in-SERP. For such queries, the brief must offer depth (examples, tools, decisions) worth clicking for, or the page should target a deeper query.

Output template

# Brief: <working title>

**Primary keyword:** <kw> (<volume>/mo, <country>) · **Variants:** <list>
**Intent:** <informational | commercial | transactional> · **Page type:** <guide | comparison | landing>
**Target URL:** <new slug or existing URL> · **Goal / CTA:** <action>
**Cannibalization check:** <none | consolidate X into this>

## SERP summary
- Format consensus: <e.g. 7/10 step-by-step guides with screenshots>
- SERP features: <AI Overview, PAA, video, snippet type>
- What top results miss: <gaps>

## Angle and information gain
<one sentence>. Evidence: <data/test/screenshots/expert>.

## Audience and E-E-A-T
Reader: <who, what they already know>. Author: <name, relevant experience>. Reviewer: <if YMYL>.

## Outline
H1: <...>
- Intro: <direct answer / promise in 2–3 sentences>
- H2: <...> — point, evidence, entities, questions, visual
  - H3: <...>
- H2: <...>
- Closing CTA: <...>

## Must cover
Entities: <list> · Questions: <PAA and forum questions mapped to sections>

## On-page
Title (≤60 chars): <...> · Meta description (140–160): <...> · Slug: <...>
Internal links out: <anchor → URL> · Internal links in: <source URL → anchor>
Schema: <types>

## Do not
<off-intent sections, claims we cannot support, competitor names to avoid>

About this skill

Written and maintained by the Claude Skills editors to the same editorial standard as every skill in the directory: a selective trigger, a real workflow, known pitfalls and a verification step. First published , last updated . MIT licensed.

esc