SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsyao-geo-panorama-audit

yao-geo-panorama-audit

General & other

View repository

Updated 10-16 08:08

About this skill

Yao GEO Panorama Audit

Use cases

  • Project initiation, quarterly review, competitor catch-up, pre-campaign evaluation, or when a GEO baseline needs to be established first.
  • Users need to first crawl public information from the target brand's official website, cross-verify brand entities, products, pages, content assets, competitors, and external sources, then output on-site and off-site system plans.
  • Output an opportunity map and P0 priorities for brand leads, growth leads, GEO consultants, and content leads.

Inapplicable scenarios

  • Diagnosing only one URL or page structure; a page diagnosis skill should be used instead.
  • Only writing rankings, comparisons, explainers, titles, or revising old articles; a content production skill should be used instead.
  • Only needing answer performance and trend monitoring on third-party Q&A platforms; a monitoring skill should be used instead.
  • Only designing GEO performance tracking, attribution fields, or data dashboards; yao-geo-tracking。
  • Already having diagnostic conclusions and only needing a 30/60/90-day execution plan broken down; an implementation plan skill should be used instead.

Required inputs

  • Brand name, official website, product lines, industry, region, target users, and business goals.
  • Known competitors, existing content inventory, core pages, brand materials, sales materials, and PR materials.
  • Seed questions, core keywords, customer inquiry records, historical SEO keyword library, and on-site search data.
  • Optional: GSC, GA, Baidu Analytics, CMS, CRM leads, third-party public opinion, and media releases.

Language and naming constraints

  • Skill names must comply with yao-geo-xxx specifications; this skill is fixed as yao-geo-panorama-audit。
  • The default output language is Simplified Chinese; the report body, table headers, field names, status values, judgment labels, and module names should all use Simplified Chinese.
  • Preserve brand names, product names, platform names, URLs, proper nouns, and users' English materials to avoid mistranslation.
  • If the user explicitly requests a bilingual or English version, first complete the Simplified Chinese main version, then generate the corresponding language version.

Required reading

  • references/skill-method.md
  • references/research-foundation.md
  • references/official-site-evidence.md
  • references/geo-quality-model.md
  • references/quality-gates.md
  • references/artifact-layout.md

Execution process

  1. First follow research-foundation.md to establish the diagnostic scope: the GEO full-scope diagnosis starts from the official website as the source of facts, with official website crawling, assertion-level verification, public source cross-verification, and on-site/off-site opportunity mapping as the main line.
  2. Crawl and archive the target brand's official website and official entry points: homepage, product pages, solution pages, pricing pages, case study pages, documentation pages, help center, blog, press releases, about page, contact page, and sitemap.
  3. Establish a brand entity profile, unifying brand names, aliases, products, services, qualifications, customers, cases, pricing, channels, regions, applicable scenarios, and inapplicable boundaries.
  4. Cross-verify official website assertions: prioritize verification with official secondary sources, help documentation, investor materials, public customer cases, media reports, encyclopedias, communities, industry reports, and authorized materials; mark content that cannot be verified as pending confirmation.
  5. Generate a user question coverage matrix, covering recommendations, comparisons, alternatives, tutorials, pricing, risks, authenticity, purchase decisions, and scenario solutions, used to check whether the official website can answer these questions.
  6. Complete the 12 diagnostic domains, see official-site-evidence.md。
  7. Score according to the unified GEO feature model: semantic density, structural standardization, citability, authority signals, readability, robustness, novelty, cross-domain contribution, and add a completeness score.
  8. Output an opportunity map, which must be split into on-site solutions and off-site solutions: on-site includes page restructuring, fact cards, FAQ, structured summaries, internal links, Schema, and content completion; off-site includes media, encyclopedias, WeChat Official Accounts/Video Accounts, communities, industry reports, partners, case evidence, and third-party endorsements.

Output contract

  • GEO 全景诊断报告
  • 方法依据与评分口径
  • 权威参考映射
  • 官网抓取与事实交叉验证表
  • 官网抓取覆盖、实体档案、产品服务与商业信息地图
  • 用户问题覆盖矩阵
  • 技术可抓取与结构化数据诊断
  • GEO 机会地图与优先级矩阵
  • 站内系统方案
  • 站外证据建设方案
  • 风险、假设与待确认项
  • 页面与内容资产修复清单
  • Default four-format delivery: Word (.docx）、PDF（.pdf）、HTML（.html or a complete HTML package), Markdown (.md）。

Checklist

  • When generating examples or formal reports, Word, PDF, HTML, and Markdown must actually be produced in four formats, and they must share the same content structure.
  • After the four formats are generated, self-review must be performed according to references/artifact-layout.md and references/quality-gates.md ; if layout overflow, border misalignment, abnormal line spacing, unreadable tables, or English-language fields are found, fix them yourself before delivery.
  • Word tables must explicitly control total width, grid column widths, and cell widths; dense tables need to be split into narrow tables and must not overflow to the right or stick to the edge.
  • PDF must be rendered from validated HTML, checking page size, pagination, table borders, and whether browser default footers exist.
  • HTML must include a fixed follow-along chapter menu bar that does not obscure content when scrolling.
  • Do not directly equate SEO rankings, official website traffic, or brand awareness with GEO priority.
  • Do not directly equate "the official website has a page" with "AI can accurately cite it"; you must check whether the page has extractable definitions, tables, FAQ, sources, update times, and assertion-level facts.
  • Do not directly equate the number of citation links with credibility; you must check whether the citations truly support the corresponding assertions.
  • Brand facts, customer cases, data, qualifications, and prices must have sources; if they cannot be confirmed, mark them as pending confirmation.
  • Conclusions must distinguish factual issues, structural issues, source issues, and execution issues.
  • The opportunity map must have resource constraints, P0/P1 priorities, recommended owners, and acceptance criteria.
  • The report must be systematic, detailed, and complete: closed-loop, assertion-level evidence, covering 12 diagnostic domains.

Reference map

  • templates/brief-template.md: standard input brief.
  • evals/trigger_cases.json: trigger and adjacent scenario tests.
  • evals/expected_artifacts.json: output contract and required files.
  • reports/output-risk-profile.md: output risk profile.
  • reports/artifact-design-profile.md: report design profile.