yao-geo-brand-graph

Yao GEO Brand Graph

Execution flow

1. By references/skill-method.md establish a source ledger, distinguishing official websites, official documentation, investor relations, customer cases, third-party materials, and materials pending confirmation.
2. By references/real-data-acquisition.md perform automated connectivity verification on source URLs; accessible sources enter the real data sample, while restricted sources enter manual review.
3. Perform entity disambiguation by mention -> candidate -> canonical entity.
4. By references/entity-schema.md extract brands, products, services, features, technologies, industries, users, scenarios, customers, cases, evidence, locations, and times.
5. By references/evidence-policy.md extract directional relationship edges; each relationship must have an evidence ID.
6. First perform a completeness self-check: authoritative references, source coverage, real data source verification, entity coverage, relationship evidence, Schema consistency, content asset alignment, domestic AI platform monitoring loop, and privacy authorization gaps.
7. Output the entity list, relationship list, credibility levels, disambiguation table, Mermaid, JSON-LD, RDF triples, domestic AI platform test scenarios, and reinforcement suggestions.

Layout quality gate

• Word must use landscape pages, real w:tbl tables, and fixed dxa table widths; automatic table widths that overflow the page are prohibited.
• The total grid width of Word tables must be less than the body width, with a safe right margin preserved.
• URLs, English entity IDs, English product names, and long English phrases must actively wrap within cells.
• HTML/PDF is laid out in the kami paper long-doc style: warm rice paper background, ivory content surface, ink blue accents, warm gray borders, serif headings, and a compact long-document rhythm.
• HTML visual reports must have a fixed following table of contents menu; the PDF print version may hide the menu to preserve body layout.
• Reports must include systematic analysis modules and must not output only basic entity tables and graph diagrams.

Tool entry points

• scripts/render_yao_geo_brand_graph.py: from structured report_input.json generate Word, PDF, HTML, Markdown, and quality-report.json。
• scripts/collect_source_validation.py: read the source ledger, automatically sample URL reachability, page titles, and verification times, and can write back to source_validation。
• examples/hubspot-domestic-ai-test/report_input.json: HubSpot domestic AI platform test sample input.
• templates/brief-template.md: project input brief template.
• evals/expected_artifacts.json: checklist for required files and sample deliverables.
• reports/output-risk-profile.md and reports/artifact-design-profile.md: output risk and layout design constraints.