seo-research
Skill6 starsChanged 3 months agoApache-2.0View on GitHub
Research conventional keywords, search intent, audience questions, demand provenance, query clusters, coverage gaps, cannibalization and competitor evidence. Map each need to one canonical page and its exact Markdown answer destination; preserve source dates, uncertainty and both opportunity tracks. Route formal multi-engine AI evidence to ai-search-research.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- empowerment-ai/agent-site-kittheir own version, first seen Sep 30, 2026
- mrieck/seoblog-claude-plugintheir own version, first seen Sep 24, 2026
- hyperfx-ai/marketing-skillstheir own version, first seen Jun 3, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- hyperfx-ai/marketing-skills★ 112same name
- empowerment-ai/agent-site-kit★ 0same name
- mrieck/seoblog-claude-plugin★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository27 other files it gives its agents.
- llms.txtllms.txt
- ai-search-researchSkill
- ai-visibility-monitorSkill
- optimise-seoSkill
- seo-action-planSkill
- seo-aeoSkill
- seo-agenticSkill
- seo-architectureSkill
- seo-auditSkill
- seo-authoritySkill
- seo-commerceSkill
- seo-competitor-pagesSkill
- seo-contentSkill
- seo-geoSkill
- seo-hreflangSkill
- seo-imagesSkill
- seo-localSkill
- seo-news-discoverSkill
- seo-pageSkill
- seo-performanceSkill
- seo-planSkill
- seo-programmaticSkill
- seo-schemaSkill
- seo-sitemapSkill
- seoSkill
- seo-technicalSkill
- seo-videoSkill
The file
We split this skill into the rules it gives an agent
11 rules. 0 are common, 8 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 10
Read markdown-delivery-contract.md for every task.
- Very specific: line 19
Read references/research-evidence-protocol.md before presenting research conclusions.
- Distinctive: line 20
Read keyword-research-protocol.md for keyword discovery, intent clustering, demand provenance, ownership, gaps, cannibalization and …
- Distinctive: line 21
Read query-opportunity-protocol.md when the goal is to choose what the site should rank for or identify the strongest next query …
- Very specific: line 27
Use ai-search-research for engine/surface-specific AI evidence, ground truth, citations, or formal Research Pack provenance.
- Distinctive: line 28
Use this skill for content/search intent, page inventory, query language, competitor/page observations, and source-backed opportunity …
- Distinctive: line 37
Research external evidence responsibly: Use primary docs, authoritative sources, and accessible pages.
- Distinctive: line 44
Do not present a snapshot of results, snippets, third-party tool data, or AI answers as stable truth.
- Distinctive: line 45
Do not scrape protected sources, bypass controls, or reproduce copyrighted source text beyond what is needed to substantiate a finding.
- Distinctive: line 46
Do not promise rankings, traffic, citations, or business outcomes.
- Distinctive: line 47
Do not convert keyword volume, page count, or competitor activity into a universal opportunity score.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
SEO Research
Markdown delivery contract
Read markdown-delivery-contract.md for every task. Write a substantive .md document for every material work step and register its direct destination in the work index.md. For public-site work, apply the owner-required Markdown discovery profile across the full eligible inventory; preserve native standards and immutable evidence formats.
Lane outputs: keyword-research.md; keyword-map.md; content-gap.md; keyword-decisions.md (combine small related steps with stable section links). Read keyword-research-protocol.md. Map intent clusters and query IDs to one canonical HTML owner and its actual Markdown destination/answer section, with demand provenance, dated SERPs, near-term and leadership tracks, gaps and cannibalization decisions.
Purpose
Build a decision-ready research note for traditional/web-search content and site decisions. This skill is for scoped discovery, not for a fabricated keyword spreadsheet, mass scraping, or a substitute for the immutable multi-engine ai-search-research Research Pack.
Read references/research-evidence-protocol.md before presenting research conclusions.
Read keyword-research-protocol.md for keyword discovery, intent clustering, demand provenance, ownership, gaps, cannibalization and Markdown answer mapping.
Read query-opportunity-protocol.md when the goal is to choose what the site should rank for or identify the strongest next query opportunity.
Input gate
Clarify audience, market/locale, decision to be made, known site/entity URLs, time horizon, access/permissions, and available first-party data. Decide whether the question is conventional search research or AI-search research:
- Use
ai-search-researchfor engine/surface-specific AI evidence, ground truth, citations, or formal Research Pack provenance. - Use this skill for content/search intent, page inventory, query language, competitor/page observations, and source-backed opportunity framing without that formal contract.
Workflow
- Define the decision. State the audience task, market/locale, entity/page scope, non-goals, and evidence required to decide.
- Build a query/question corpus. Separate user needs, observed search queries, AI prompts, and engine-executed subqueries. Group language by task and stage, not just lexical similarity. Preserve parent families, source, date, locale, country/location, device, engine/surface, conversation turn, coverage, confidence, and limitations. Hash-pin local source and coverage evidence. Treat volume/difficulty/vendor metrics as dated estimates, never ground truth.
- Inspect existing coverage and performance. Map actual pages/assets to the questions they serve. When authorized first-party data exists, preserve query, page, country, device, search type, time window, clicks, impressions, CTR, and average position together; identify head-term/category leadership, near-win visibility, snippet/CTR, coverage-gap, defend, and cannibalization candidates without treating average position as an exact rank. Absence from first-party rows does not remove an owner-mandated high-volume family; retain it with the demand/SERP evidence and record the visibility gap.
- Inspect the current result set. For candidate query families, record a dated locale/device SERP observation: dominant intent, result/page types, visible title/snippet patterns, strong competitor evidence, freshness, authority expectations, and features that change the reader task. A SERP snapshot is evidence for fit, not a stable ranking formula.
- Select page-level query ownership. Choose one primary need/query family and natural supporting language per target page. Preserve owner-named strategic queries, business value, achievable page fit, existing authority, coverage state, and evidence gaps. The highest-volume phrase is not automatically the best short-term opportunity, but every owner-declared relevant high-volume family and the market's highest-relevant-demand family must receive an explicit head-term leadership decision. Difficulty changes the route; it does not erase the ambition.
- Research external evidence responsibly. Use primary docs, authoritative sources, and accessible pages. Capture title, URL, access date, source type, claim, and limitations. Respect robots, terms, rate limits, paywalls, authentication, and copyright.
- Synthesize choices. Return a two-track opportunity slate: nearer-term capture/defense work and high-volume head-term/category leadership. Use
now,next, andlaterto sequence both tracks without allowing easy work to replace the leadership target. Usedeferfor a documented owner decision, genuine intent/product mismatch, or prohibited/misleading target—not merely high difficulty, weak authority, cost, or current absence. Separate observed facts, inferred opportunities, experiments, and unknowns; do not collapse unlike dimensions into a universal keyword score. - Freeze formal query work. Let
<suite-root>mean${CLAUDE_PLUGIN_ROOT}in Claude Code. In Codex, read.seo-suite-runtime.jsonbeside thisSKILL.mdwhen present and use itssuite_rootvalue; otherwise use the absolute repository checkout. When the corpus will drive a baseline, audit, page map, or implementation decision, createquery-corpus.jsonusing the checked-out suite contract and validate it withpython "<suite-root>/scripts/validate_query_corpus.py" validate-corpus <bundle>/query-corpus.json --bundle <bundle>. - Hand off. Send conventional performance measurement to
seo-performance, page search-result review toseo-page, content work toseo-content, competitor pages toseo-competitor-pages, implementation sequencing toseo-action-plan, and AI-specific formal work toai-search-research.
Guardrails
- Do not present a snapshot of results, snippets, third-party tool data, or AI answers as stable truth.
- Do not scrape protected sources, bypass controls, or reproduce copyrighted source text beyond what is needed to substantiate a finding.
- Do not promise rankings, traffic, citations, or business outcomes.
- Do not convert keyword volume, page count, or competitor activity into a universal opportunity score.
Formal evidence handoff
When this work needs a cross-team, approval-ready plan, package evidence-bound findings as an immutable seo-findings.json bundle using the checked-out suite contract. Keep every referenced capture/source below raw/, retain declined claims and limitations, and run python "<suite-root>/scripts/validate_seo_findings.py" validate-findings <bundle>/seo-findings.json --bundle <bundle>. Send only a passing bundle to seo-action-plan; otherwise label the handoff provisional.
Output
Return a decision statement, validated query corpus when required, source register, existing-coverage and query-to-page map, current first-party performance where available, dated SERP intent review, the mandatory head-term leadership track, nearer-term opportunity track, selected primary/supporting query families, leading-result gap map, staged work and measurement milestones, observation/inference split, declined ideas, owners, and evidence gaps. Include collection dates and a refresh trigger for volatile sources.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

