search-ready · diff
git:20260807.975fe85 to git:20260807.fac9f85
47 added, 22 removed. Audit A to A.
---
name: search-ready
- description: Audit and improve technical SEO, content, structured data, social previews, and AI-search visibility. Use when preparing or maintaining a site for discovery.
+ description: Audit, implement, or verify technical SEO, indexation, content, schema, social previews, and AI discovery. Use when preparing or maintaining a live site for search and answer engines.
license: MIT
---
# Search readiness
- Make a real site easier to crawl, understand, rank, cite, and share. Work from live evidence and the repository. Do not produce a generic checklist detached from the implementation.
+ Make a real site easier to crawl, index, understand, serve, cite, and share. Work from live behavior, repository intent, and search data as separate evidence classes.
+ ## Choose a mode
+
+ - **Audit:** diagnose live and repository issues without changing files.
+ - **Implement:** apply approved fixes and verify deployment behavior.
+ - **Content opportunity:** research current queries, competitors, and audience questions.
+ - **Structured data:** select, implement, and validate eligible markup.
+ - **AI discovery:** evaluate crawler policy, answer-ready content, entity clarity, and experimental aids.
+ - **Maintenance:** compare current state with prior evidence and define recurring checks.
+
+ Route by site type when relevant: SaaS/docs, editorial, commerce, local, or international. Do not use an equal-weight generic score across unlike sites.
+
## Workflow
- 1. Confirm the canonical production URL, important routes, audience, conversion action, and access to the codebase.
- 2. Fetch the live site and inspect the rendered HTML, response headers, redirects, robots rules, sitemap, canonicals, metadata, structured data, internal links, and status codes.
- 3. Search the web for branded queries, category queries, direct competitors, current search-result formats, and sources that answer the audience's questions. Treat rankings and snippets as time-sensitive evidence.
- 4. Inventory indexable routes and assign each a search job. Identify duplicates, thin pages, orphaned pages, accidental noindex rules, and missing content.
- 5. Fix technical blockers before content polish: crawlability, canonicalization, metadata, schema validity, rendering, performance, and sitemap coverage.
- 6. Improve page content around real questions and claims. Put the direct answer near the top, keep terminology consistent, cite primary sources, and add proof the project can support.
- 7. Add AI-discovery support where useful: explicit entity descriptions, answer-shaped sections, accessible HTML, crawler policy, and `llms.txt`. Never present `llms.txt` as a ranking guarantee.
- 8. Verify social previews for important routes. If the project needs new preview artwork or dynamic images, follow the `social-sharing` skill.
- 9. Re-fetch the deployed result, validate structured data, compare metadata route by route, and report what changed, what remains blocked, and what should be checked monthly.
+ 1. Confirm canonical production host, important routes/templates, audience, conversion action, locales, site type, and repository access. Inspect available Search Console/Bing, analytics, logs, prior audits, and deployment configuration before asking questions.
+ 2. Fetch representative live URLs and inspect rendered HTML, source HTML where useful, headers, redirects, robots, sitemap/indexes, canonicals, hreflang, metadata, structured data, internal links, status behavior, and rendering dependencies.
+ 3. Compare live behavior with repository intent. Label observed live behavior, source intent, search-platform data, and recommendation separately.
+ 4. Inventory indexable route classes and assign each a search job. Diagnose duplicates, soft 404s, thin/orphaned pages, faceted/paginated states, accidental noindex, canonical conflicts, redirect chains, and sitemap coverage.
+ 5. Prioritize blockers by likely impact on crawl, index, serve, comprehension, user experience, and business value—not character counts or checklist totals.
+ 6. Research current result formats, branded/category queries, competitors, and source ecosystems when the task needs content or citation evidence. Treat rankings and snippets as time-sensitive.
+ 7. Improve content around real questions and defensible claims. Preserve natural structure; do not force question headings, FAQs, word counts, or statistics at arbitrary intervals.
+ 8. Add structured data only when it matches visible content, applicable schema, and current search-feature policies. Validate syntax and eligibility separately.
+ 9. Treat AI-search practices by confidence: provider-documented crawler controls, standard accessible content, experimental conventions such as `llms.txt`, and unsupported speculation. Never promise rankings or citations.
+ 10. Re-fetch the deployed result, run relevant validators and performance checks, compare route samples, and report remaining blockers and maintenance needs.
- ## Load deeper guidance
+ ## Load conditional references
- - Read [technical SEO](references/technical-seo.md) for the repository audit, implementation patterns, scoring model, and verification commands.
- - Read [AI-search workflow](references/ai-search-workflow.md) when the task includes answer engines, citations, authority building, or current web research.
- - Read [AI-search reference](references/ai-search-reference.md) for schema, crawler, directory, and outreach details.
+ - Read [technical SEO](references/technical-seo.md) for route sampling, diagnostics, implementation patterns, and validation.
+ - Read [AI-search workflow](references/ai-search-workflow.md) only for answer engines, citations, crawler policy, or entity/content work.
+ - Read [AI-search reference](references/ai-search-reference.md) only for conditional schema, crawler, `llms.txt`, and outreach patterns.
- ## Output
+ This file controls prioritization, safety, output, and verification.
- Return:
+ ## Output contract
- - a prioritized findings table with evidence, affected URLs, severity, and fix;
+ Return mode-relevant sections:
+
+ - scope and evidence sources;
+ - prioritized findings with affected URLs/templates, severity, confidence, and fix;
+ - live-vs-repository discrepancies;
+ - route-level metadata/schema/indexation matrix;
- implemented changes with file paths;
- - a route-level metadata and schema matrix;
- - content opportunities grounded in live queries and competitors;
- - verification results and a short maintenance cadence.
+ - content opportunities grounded in current evidence;
+ - validator and deployed verification results;
+ - unresolved blockers and maintenance cadence.
- Do not invent traffic, rankings, authority, or indexation. Distinguish observed live behavior from repository intent and from recommendations.
+ ## Verify
+
+ - Representative templates and edge states were sampled, not just the homepage.
+ - Live response, rendered content, and repository intent are distinguished.
+ - Indexation claims use search-platform evidence when available and are otherwise qualified.
+ - Schema matches visible content and passes relevant validation.
+ - Redirects, canonicals, robots, sitemap, hreflang, status codes, and rendering agree.
+ - Performance claims identify lab versus field data.
+ - AI-search recommendations state evidence confidence and do not treat experimental conventions as guarantees.