Technical SEO
Use this skill to keep search-facing work canonical: one public URL, one copy source, one indexing inventory, and one proof path for each SEO claim.
Workflow
- Name the SEO branch:
- Metadata copy.
- Canonical URL or redirect.
- Sitemap or robots.
- Structured data.
- Public image alt text or generated HTML.
- Dynamic public inventory.
- Indexing submission or inspection.
- Crawlability, SSR payload, or rendering performance.
- Completion criterion: the affected public routes and SEO branch are explicit.
- Inspect the current source of truth:
- Search route files, copy constants, head helpers, sitemap, robots, redirects, footer/nav links, generated routes, and SEO tests.
- If the task depends on search-engine behavior or indexing APIs, verify current official docs or webmaster tooling before acting.
- Completion criterion: every assumption that affects SEO behavior is verified in current source or marked unknown.
- Preserve copy authority:
- Treat user-supplied SEO, legal, pricing, product, or landing copy as exact unless the user asks for a rewrite.
- Prefer an existing shared copy object or route contract over route-local duplicate strings.
- Check truncation rules before changing descriptions.
- Completion criterion: the authoritative copy source and route consumer are named.
- Make the route canonical:
- Pick the canonical public URL.
- Keep old URLs only as intentional redirects or compatibility aliases.
- Sweep footer links, nav links, sitemap entries, internal rich-text links, runtime allowlists, and tests for stale paths.
- Completion criterion: old-path and new-path searches have been reviewed, and every remaining old path has a reason.
- Align metadata:
- Emit title, description, canonical link, robots meta, Open Graph, Twitter card, and primary image consistently.
- Use
noindexfor private, draft, unavailable, duplicate, admin, auth-only, or non-ready pages. - Add structured data only when the page content actually supports it.
- Completion criterion: metadata matches the page's real public content and indexability state.
- Align indexing inventory:
- Ensure
robots.txtallows intended crawlers and blocks private or non-public surfaces. - Ensure
sitemap.xmlcontains only canonical, indexable public URLs. - Include
lastmodonly when the source timestamp is meaningful and formatted correctly. - Dynamic sitemap rows must be filtered by public visibility and readiness.
- Completion criterion: sitemap and robots behavior match the canonical route policy.
- Ensure
- Verify submission separately from indexing:
- For IndexNow or similar push APIs, verify key-file reachability and request acceptance.
- For Google, use sitemap submission or Search Console inspection according to the current supported workflow.
- Never say a URL is indexed unless an inspection tool or search-engine source proves it.
- Completion criterion: the report distinguishes submitted, accepted, crawled, and indexed.
- Add guardrails:
- Update focused tests for copy exactness, canonical paths, sitemap inventory, robots rules, redirects, structured data, and image alt text.
- Regenerate route manifests after route-file changes.
- Run the narrowest useful checks first, then broader checks when shared behavior changed.
- Completion criterion: verification commands and residual risks are reported.
Output Rules
- Lead with what changed and what was verified.
- State stale assumptions explicitly.
- Report indexing claims precisely:
submitted,accepted,inspected, orindexed. - If verification is blocked, name the missing tool, credential, live URL, or external source.
Traps
- A route can be fixed while footer, sitemap, nav, generated routes, or tests still point to an old URL.
- Exact copy can be silently truncated by a shared metadata helper.
- A successful build does not prove crawler-visible HTML, canonical redirects, or indexing submission.
- Indexing submission acceptance is not indexing.
- Search-engine APIs and webmaster rules change, so verify current official behavior when it matters.