Note

This README was generated by SKILL, get the ZH version from here.
This skill's implementation was fully agent-generated; the developer only tuned its input/output.

---

SEO AND AEO GROUNDED IN THIS SEASON'S RESEARCH!

---

An agent skill with mandatory fresh research, surface-aware optimization rules, and a plan-before-apply workflow

Table of Contents

• Features
• Architecture
• License

Features

/seo-optimize [PROJECT_PATH] [--plan] [--reset] · Documentation

• Mandatory Research on Every Run — Eleven parallel queries plus live WebFetch of Google, OpenAI, and Anthropic primary docs run each time, and the verified-position snapshot is rewritten in place when it disagrees, so tactics that reversed within six months never get applied from memory.
• Primary Sources Override Industry Myths — Sources are ranked in four tiers where official docs overrule blogs and AEO SaaS guides, overturned claims land in a "myths" column instead of the action list, and existing dead weight such as an llms.txt on a non-agent site is flagged as removable.
• Surface-Aware Rule Routing — The analyzer inventories code type and deployable web surfaces separately, so a library or CLI without a site only gets registry, GitHub, and entity-consistency work rather than fabricated page fixes.
• Training vs. Retrieval Crawler Split — robots.txt rules are classified by bot purpose, a wildcard site-wide block or a blocked retrieval bot is marked Critical, and opting out of training never costs citation eligibility.
• Plan-Then-Apply Confirmation Gate — A plan file with rule IDs and diff previews is produced and confirmed before any file changes, remote gh repo edit commands are listed but never run, and no git writes occur.

Architecture

Full Architecture

graph TB
    User[User] -->|/seo-optimize| Skill[SKILL.md<br/>Orchestration]
    Skill --> Research[Research Protocol<br/>Phase A / B]
    Research --> Anchors[knowledge_anchors.md<br/>Position Snapshot]
    Skill --> Analyze[analyze_seo.py<br/>Surface Inventory]
    Skill --> Ask[Keyword Prompt<br/>config.json]
    Analyze --> Rules[optimization_rules.md<br/>R1–R12 Routing]
    Research --> Rules
    Ask --> Rules
    Rules --> Plan[Plan File<br/>Await Confirmation]
    Plan --> Apply[Apply + Result Report]

Loading

License

This project is licensed under the MIT LICENSE.