yaojingang/yao-geo-skills

Yao GEO Skills – What It Is

Yao GEO Skills is an open‑source collection of “skills” (reusable, version‑controlled packages) that help teams automate and standardise the workflow around Generative Engine Optimization (GEO) – a term the authors use for improving how content and pages are discovered, extracted and ranked by AI‑powered search engines (e.g., DeepSeek, Doubao, Kimi, ChatGPT, etc.).

Why it matters

• GEO is a niche but growing discipline that sits at the intersection of SEO, prompt engineering, and AI‑generated content. Companies that rely on AI‑driven search need repeatable processes for:
  • Auditing pages for AI‑readability
  • Building brand knowledge graphs
  • Generating AI‑friendly titles, articles, comparison tables, and ranking lists
  • Tracking the downstream impact of those assets on AI answer generation.
• The repo bundles the whole pipeline – from a clear SKILL.md entry point to validation scripts, templates, and example outputs – so teams can reuse the same logic without recreating prompts from scratch.

Core concepts

Concept	What it is	Typical output
Skill	A self‑contained package (manifest, templates, eval data, execution script) that performs a single GEO‑related task.	Markdown, HTML, DOCX, PDF, JSON, Mermaid diagrams, etc.
SKILL.md	Human‑readable description that defines the agent’s entry point, input boundaries and expected artefacts.	Documentation for users and CI checks
manifest.json	Machine‑readable metadata (name, category, inputs, outputs).	Enables repository‑wide validation
evals/	Test cases (trigger_cases.json) and expected results (expected_artifacts.json) to guarantee reproducibility.	Automated CI validation
agents/interface.yaml	Stable interface definition that downstream tools can call.	Consistent API across all skills

What you get out of the box

• 24 ready‑made skills covering the full GEO lifecycle:
  • Operations – CLI integration with the GEOFlow platform, template migration, Laravel Blade theme editing.
  • Strategy – Panorama audits and 30/60/90‑day execution roadmaps.
  • Page‑technical – Page audits and blueprint generation for AI‑friendly markup.
  • Content‑production – Title optimisation, comparison reports, ranking‑article builders, explainer writers, content refiners, article‑friendly converters.
  • Knowledge‑assets – Brand graph construction and knowledge‑base generation.
  • Measurement – Tracking plans and AI‑answer monitoring dashboards.
  • Research – Prompt collections, keyword mining from Baidu, Xiaohongshu, Douyin, etc.
• Example artefacts for every skill (Markdown, HTML, Word, PDF, quality‑report JSON) so you can see exactly what the skill should produce.
• Validation script (scripts/validate_repository.py) that checks the required file structure and ensures each skill passes its own test cases.

How to use it

1. Clone the repo (or sparse‑checkout only the skills you need).
2. Run the validation script to confirm the repository is well‑formed.
3. Follow the SKILL.md of a chosen skill to supply the required brief/template input.
4. Execute the skill via the defined agent interface (often a CLI or Python entry point).
5. Inspect the generated artefacts – ready to be handed to designers, writers, or directly uploaded to a CMS.

Who should consider it

• Marketing or growth teams that run AI‑driven content campaigns and need repeatable, auditable processes.
• SEO/AI‑search specialists building brand knowledge graphs for large language models.
• Product teams that already use the GEOFlow platform and want plug‑in automation.
• Anyone looking for a structured, open‑source alternative to ad‑hoc prompt collections.

Limitations / Not a fit for

• Pure prompt‑library collections – this repo expects full‑stack, validated skill packages.
• Projects that rely on private APIs or internal data that cannot be stripped for open‑source release.
• General‑purpose AI tooling unrelated to GEO (e.g., image generation, robotics, LLM training).

---

All details above are taken directly from the repository’s README; no external assumptions have been added.