Free Claude Code SEO Skill Blueprint

Copy-ready skill sections

• Purpose: Convert one live scan finding into one bounded, reviewable repository change.
• Required inputs: Production URL, affected route, observed and expected behavior, evidence, boundaries, and validation commands.
• Discovery: Read instructions and inspect routes, renderers, content, metadata, sitemaps, policies, styles, and tests.
• Implementation: Fix the highest dependency with the smallest safe change and truthful content.
• Validation: Run file-specific checks, resolve links, inspect rendered output, schema, and sitemap coverage.
• Handoff: Separate locally proven work from deployment, production response, indexing, and traffic evidence.

Main Explanation

Start the skill with a narrow purpose: convert one observable public website issue into one safe repository change and a verification report. The description should mention scan-backed SEO implementation so it does not load for unrelated coding tasks. Follow the current official Claude Code skill format rather than copying an old directory convention from a third-party post.

Define required inputs before the workflow begins: canonical production URL, affected route, observed behavior, expected behavior, supporting scan evidence, known platform or framework, public versus private route boundary, and normal validation commands. If the input only says improve SEO, the skill should request or collect evidence instead of inventing a task.

The first workflow phase is discovery. Read project instructions completely. Locate route resolution, shared layouts, metadata, content sources, sitemap logic, robots controls, structured-data rendering, styles, and relevant tests. Search the content catalog for overlap before adding a URL. Report the likely edit surface and dependencies before changing files.

The second phase is implementation. Fix the highest dependency first and make the smallest correct change. Preserve authentication, authorization, billing, checkout, webhooks, analytics, user data, private dashboards, APIs, and unrelated design. Never create fake reviews, customers, performance numbers, authors, dates, prices, or integrations. Never expose a private route to make a scan pass.

The third phase is content quality. Give each public page one visitor job. Lead with a direct answer, explain when it applies, include limitations and a verification method, and connect the page to useful hubs and siblings. Consolidate keyword variants that share intent. Match visible FAQ content exactly to structured data.

The fourth phase is local verification. Run syntax checks and tests proportional to the touched files. Resolve every new internal link. Render the page when layout or responsive behavior changed. Inspect one H1, heading order, canonical, robots meta, Markdown alternate, image path, FAQ schema, and sitemap entries. Review the final diff for accidental unrelated edits.

The fifth phase is handoff. State which evidence was observed, which files changed, what local checks passed, and what remains unproven. Do not mark crawling, indexing, ranking, AI visibility, citations, or traffic as complete. Provide the production route checks and rescan steps the owner must run after deployment.

Keep deterministic enforcement outside the prose workflow. Use permissions or a pre-tool hook to block sensitive operations, and a post-edit hook for mandatory formatting or linting. The skill can explain how to respond to failures, but an interpreted instruction is not the same as technical enforcement.

Practical Steps

• Create a project skill using the current Claude Code documentation.
• Use the narrow audit-to-fix purpose and trigger.
• Add the six blueprint sections from this page.
• Replace generic validation with the repository’s real commands.
• List protected files, routes, data, and external actions.
• Run TruboRankAI and supply one real finding.
• Review the proposed scope before allowing edits.
• Deploy, inspect the public route, and rescan.

FAQ

Related internal links