seo-setup
/seo-setup
Ustawia SEO strony: rejestruje ją w Search Console i Bing, dodaje sitemap, robots.txt i metadane, a potem zapisuje listę zadań w TODO.md.
- Autor
- workbench-soft ↗
- Repozytorium
- agent-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 5
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 2
Brak sygnałów. Repozytorium zmienione 13.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Ustawia SEO strony: rejestruje ją w Search Console i Bing, dodaje sitemap, robots.txt i metadane, a potem zapisuje listę zadań w TODO.md.
Wyzwalacze z opisu
Oryginalny opis (EN)
Use when the user asks to set up SEO, get a site indexed, register it in Search Console or Bing, add Cloudflare Web Analytics, a sitemap, robots.txt, meta tags or structured data, or prepare for AdSense review: "set up SEO", "настрой SEO", "зарегай сайт в гугле". Finds where DNS sits, checks the live site and repo against a list, fixes the code, goes through the dashboards with the user logged in, writes TODO.md.
Jak działa
- Logins and keys
- 1. Read the project
- 2. Find where the domain lives
- 3. Audit the site
- 4. Fix the gaps in the repo
- 5. Plan and write content
- 6. Register the site
- 7. Leave a record
- 8. Write TODO.md
- 9. Report back
Podgląd SKILL.md
--- name: seo-setup description: 'Use when the user asks to set up SEO, get a site indexed, register it in Search Console or Bing, add Cloudflare Web Analytics, a sitemap, robots.txt, meta tags or structured data, or prepare for AdSense review: "set up SEO", "настрой SEO", "зарегай сайт в гугле". Finds where DNS sits, checks the live site and repo against a list, fixes the code, goes through the dashboards with the user logged in, writes TODO.md.' --- # Set up search for a site The agent MUST get the site found and ready for ad review, then register it with Cloudflare, Google and Bing. ## Logins and keys - **Browser** the agent MUST drive whatever browser tool it has: the VS Code browser tools, or the `agent-browser` skill. Without one, it MUST hand the user the link and the clicks as numbered steps, one per line, and wait. - **Login screen** the user MUST log in inside that window, and the agent MUST wait for "done". A password, a 2FA code or card details MUST NOT be typed by the agent. - **Account choice** the agent MUST ask a single time which Google and Cloudflare accounts to use, then stick to them in each dashboard. - **Secrets** an API token MUST stay out of commits and out of the chat. The user stores it with `gh secret set NAME`, or the agent reads it from a file outside the repo. - **Public IDs** a verification token, a site key or an analytics ID MAY sit in the code. The page shows them to the world already. ## 1. Read the project The repo rules MUST beat this skill. The agent MUST start with `AGENTS.md`, `CLAUDE.md` and the README, then look for: 1. **The domain** the agent MUST find it in the README, a `CNAME` file, `wrangler.jsonc` or `wrangler.toml`, `vercel.json`, `netlify.toml`, or the canonical link in the HTML. 2. **The host and the deploy** the agent MUST find them in those same files and `.github/workflows/`. 3. **How the pages are made** the agent MUST name it: static HTML, a prerender step, server rendering, or an app that renders only in the client. 4. **The work already there** the agent MUST reuse it: a README section on search and analytics, a sitemap, `robots.txt`, JSON-LD. When the site lives only on `*.pages.dev`, `*.github.io` or `*.vercel.app`, with no domain of its own, the agent MUST stop and say so. AdSense wants an own domain, and Cloudflare Registrar sells one at cost. The agent MUST continue only once the user chooses: buy a domain, or go on with no ads. ## 2. Find where the domain lives The agent MUST run these checks: ```bash d=example.com dig +short NS "$d" # *.ns.cloudflare.com: DNS on Cloudflare whois "$d" | grep -iE 'registrar:|name server' | sort -u curl -sI "https://$d" | grep -iE '^(server|cf-ray):' # server: cloudflare: proxied ``` The agent MUST take the path that matches what the checks show: | What the checks show | Path | | ---------------------------- | ---------------------------------------------------------------------------------------------------- | | Cloudflare DNS, proxied | the Cloudflare path below, every step | | Cloudflare DNS, grey cloud | ask before turning the proxy on, it moves traffic through Cloudflare | | DNS at another host | offer moving DNS to Cloudflare, free plan. The user decides, the nameservers change at the registrar | | The user keeps DNS elsewhere | the same steps, with the TXT records added at that host and the analytics snippet in the page | The Cloudflare settings are in [references/cloudflare.md](references/cloudflare.md), and the agent MUST read it. ## 3. Audit the site The agent MUST go over the live site and the repo. It MUST read [references/site-checklist.md](references/site-checklist.md) and work through each row in it. - The agent MUST put the result in a table: check, state, fix needed. - Each check MUST be taken from the live site or the build output. The source on its own misses what the host adds or strips. - About, Contact and a privacy policy are the trust pages, and they MUST exist. Both ad review and Google Play look for them. ## 4. Fix the gaps in the repo The agent MUST show the audit table and the list of fixes, then begin with no question. A fix MUST get a question first only when it changes a lot of what people see, such as a new page layout or a rename. - A fix SHOULD get a commit of its own, in the repo's commit style. - The sitemap SHOULD be built from the page list the build renders. A sitemap written by hand is out of date at the first new page. - Text for people MUST follow the `writing-style` skill. The agent MUST load the skill before the first page of text. - Before each commit, the agent MUST run the repo checks: tests, lint, build. - The agent MUST deploy the repo's usual way, then audit the live site a second time. ## 5. Plan and write content People reach a site from search through pages that answer their query. The agent MUST read [references/content.md](references/content.md) if the site has under 5 pages of real text, or if the user asks for content. - The agent MUST show the list of topics and wait for the user's pick. An unwanted page costs a day of writing. - Each number and rule in the text MUST trace to a source that the page links. ## 6. Register the site The agent MUST read [references/search-consoles.md](references/search-consoles.md), then go through each dashboard in the browser: 1. **Cloudflare Web Analytics** the agent MUST add the site. 2. **Cloudflare Crawler Hints** the agent MUST switch it on, so IndexNow tells Bing about changes. 3. **Google Search Console** the agent MUST add the domain property, verify it, send in the sitemap, request indexing of the key pages. 4. **Bing Webmaster Tools** the agent MUST import from Search Console and check the sitemap. 5. **Rich Results Test and PageSpeed** the agent MUST run both on the home page and on one page of each type. When the dashboard shows a step already done, the step SHOULD be checked, then skipped. ## 7. Leave a record In the target repo README, the agent MUST add or update a section, written per `writing-style`: what is registered where, the account that owns each, how the site got verified, the sitemap's location. Secrets and passwords MUST be left out. The skill's next run SHOULD start from that section. That spares a walk through each dashboard again. ## 8. Write TODO.md `TODO.md` at the target repo root MUST hold each task this run leaves for the user. The agent MUST touch only the `## seo-setup` section, adding or updating it. - **One task per line** each line MUST hold a checkbox, the task, the place, and a date for a wait. - **Done tasks** MUST come off once the user says so or the dashboard shows it. - **No tasks left** the section MUST be removed. - **Secrets** passwords and tokens MUST be left out. The section SHOULD read like this: ```markdown ## seo-setup - [ ] Read Pages in Search Console on <date>: https://search.google.com/search-console - [ ] Add the TXT record `google-site-verification=...` at the DNS host ``` The agent MUST commit it together with the README record. ## 9. Report back 1. The answer MUST give the audit table and the changes made. 2. It MUST list the commits, a line for each. 3. It MUST name each service: registered or not, with its dashboard link. 4. It MUST say what is waiting on Google or Bing, and when to look: Search Console data appears in 2 to 3 days. 5. It MUST list the `TODO.md` tasks, a line for each. ## Guardrails - A DNS record the agent didn't add MUST stay as it is, unchanged and in place. - A Cloudflare setting that changes the traffic path, the proxy, the SSL mode or a WAF rule, MUST get the user's agreement before it changes. - A redirect rule MUST be tested on the live site as soon as it goes in. - `noindex`, a `Disallow: /` or a robots block MUST stay off the live site.
Pliki w paczce (5)
- SKILL.md 8 KB
- references/cloudflare.md 4 KB
- references/content.md 3 KB
- references/search-consoles.md 3 KB
- references/site-checklist.md 5 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · workbench-soft/agent-skills | 100 | MIT |
| KILN-skills/kiln | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki ff96798ce0a2 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 95seo
Sprawdza SEO projektu Next.js: metadane, sitemapę, robots.txt, Open Graph, dane strukturalne i dostępność. - 100seo-launch
Przygotowuje publiczne strony pod wyszukiwarki: frazy, tytuły, sitemap, robots.txt, Search Console i sprawdzenie indeksacji przed startem. - 100better-seo-build
Konfiguruje SEO od zera w projekcie strony: robots.txt, sitemapę, llms.txt, metadane i dane JSON-LD w konwencjach frameworka. - 100seo
Przeprowadza techniczny audyt SEO pod Google, Bing i Yandex za pomocą 80 sprawdzeń i wskazuje, co naprawić, bez zmian w stronie. - 100google-search-console
Dodaje stronę do Google Search Console: weryfikuje właściciela, sprawdza mapę witryny i robots.txt, wysyła mapę i wyjaśnia raporty nieindeksowania. - 95seo
Wdraża lub audytuje metadane SEO: tagi Open Graph, dane JSON-LD, mapę witryny, robots.txt, kanoniczne adresy i podglądy w mediach społecznościowych.