사용 방법
- 설치 명령을 복사하고 자신의 에이전트 환경에 설치하기 전에 소스 패키지를 확인하세요.
- 필요한 도구, 인증 정보, 지원 환경은 원본 문서에서 확인하세요.
- 입력과 원하는 산출물을 설명하고 출처와 결과를 대조하세요.
npx skills add https://github.com/anycap-ai/anycap --skill anycap-ai-tool-seo시작 전 확인
- 설치에는 모델 구독, API 한도, 파일 접근 권한이 포함되지 않습니다.
- 별점 수와 설치 수는 활동 지표이며 독립적인 품질·보안 검증이 아닙니다.
원본 저장소의 제목, 예시, 코드, 표, 링크, 이미지를 유지해 표시합니다.
AnyCap AI Tool SEO
Audit search visibility and decide what to fix, improve, consolidate, or publish next. A technically valid page is only eligible for indexing; it is not guaranteed to be crawled, indexed, or ranked.
Read framework.md, page-planning.md, and guardrails.md before recommending site-wide changes. Use the current site and first-party evidence as the source of truth for its status.
Workflow
1. Establish the baseline
- Identify the site, searcher, target market, conversion path, current URL inventory, and recently changed or retired URLs. Infer missing context from the site when possible.
- In Google Search Console (GSC), compare complete date ranges and page/query cohorts. Separate branded from non-branded searches and current URLs from retired URLs. Record impressions, clicks, index coverage, and the last complete data date. Do not infer that a past algorithmic visibility loss has ended solely because manual actions and security issues are clear.
- Use Ahrefs or similar tools for competitor discovery and keyword estimates; use GSC for the site's observed Google Search performance. Label estimates and time windows.
- If GSC is unavailable, say which conclusions remain unverified and continue with public crawling and site evidence.
2. Map each intent to an existing URL
- Before proposing a page, identify its current owner. Check that URL's response code, robots directives, canonical, sitemap presence, internal links, rendered content, and GSC URL Inspection when available.
- Distinguish stored inspection data from a live test. Record Google's last crawl date. A live test showing eligibility does not mean the URL is indexed.
- Apply the decision table in framework.md. Fix a genuine technical blocker first; improve or consolidate an existing owner before adding a competing URL. Allow recent changes time to be recrawled before judging their effect.
3. Research demand and SERP intent
- Inspect current search results in the target country and language, preferably through a search interface that actually shows that engine's results. Record dominant intent, page types, evidence, and gaps. One query sample is not a stable ranking study.
anycap search --query "<topic>" --no-crawl --max-results 10can find public pages for further reading. It is a general web search sample, not a Google rank tracker or source of search volume. Use it only when the CLI is available and helpful; authentication is not required for the rest of this workflow.- Validate whether the product fulfills the searcher's task. Avoid keyword variants that duplicate the same intent and page owner.
4. Choose and execute the smallest useful change
- Improve an existing page when it already owns the intent: answer the task clearly, add verified first-hand evidence, state real limitations, make the product path useful, and link from relevant pages.
- Consolidate overlapping pages when they split the same intent. Create a new page only for a distinct task the current inventory does not serve.
- Use structured data only when the page and Google feature support it. Do not add FAQ sections or markup mechanically.
- For translated pages, verify that the main content is genuinely localized and that canonical and hreflang relationships are correct. Do not prune a language merely because a short GSC window has few impressions.
5. Prioritize and measure
- Rank opportunities by observed blocker or demand, user value, evidence readiness, implementation effort, and risk of duplicate intent. Fix shared crawl/index issues before scaling content.
- Publish a small, reviewable set. Track crawl/index status and page/query cohorts after Google processes the change; avoid fixed ranking deadlines or promises. Sitemap submission and recrawl requests are discovery aids, not indexing guarantees.
- Deliver a concise evidence table: query/task, current owner, GSC status/date, SERP finding, action (fix/update/consolidate/new/wait), priority, and measurement plan. Add page briefs only for justified new or substantial rewrite work.
Optional AnyCap commands
For web research when the CLI is available, consult the anycap-cli skill for syntax and current behavior:
anycap crawl https://example.com
anycap search --query "image generation api" --no-crawl --max-results 10These commands do not report Google Search Console metrics, Google index status, or exact Google SERP positions. Do not run unrelated model-generation tests as a prerequisite for SEO diagnosis.
Evidence and privacy
- Separate official search guidance, observed site data, third-party estimates, and practitioner hypotheses.
- Do not fabricate product tests, testimonials, benchmarks, customer logos, or rankings.
- Keep authenticated GSC/Ahrefs exports, account identifiers, and private dashboard screenshots in a private ignored location; publish only sanitized conclusions when a repository record is needed.
Resources
- framework.md — baseline, indexing decisions, and prioritization
- page-planning.md — intent ownership and page briefs
- guardrails.md — Google guidance and quality boundaries



