使用方法
- 复制安装命令，在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物，并对照来源检查输出。
npx skills add https://github.com/sergekostenchuk/ui-ux-agent-skill-system --skill seo-regression-validator开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况，不等于独立质量或安全测试。
仓库原始内容
按源仓库内容呈现，保留标题、案例、代码、表格、链接以及原文引用的演示图片。
SEO Regression Validator
Use this skill after technical SEO/schema guidance exists, before claiming a site or skill output is done.
Read references/regression-checklist.md before running a broad audit.
Owns
- live HTML and local fixture checks;
- pass/fail severity model;
- timestamped evidence reports;
- schema/head metadata checks;
- public artifact reachability checks;
- regression report templates;
- negative tests for missing JSON-LD, duplicate canonical, missing image, and fake reports.
Does Not Own
- implementing SEO fixes;
- Search Console credentialed data;
- analytics exports;
- external authority placement;
- mutating target websites.
Workflow
- Define target URLs or fixture files.
- Record timestamp, target, method, and user agent.
- Fetch or read the target without mutation.
- Run deterministic checks using scripts/audit_static_seo.py when HTML is available.
- Separate critical failures, warnings, and skipped checks.
- Record skipped checks instead of treating them as passes.
- Produce or update
seo-regression-report.jsonand a human-readable validation summary. - Route implementation fixes to the appropriate specialist.
Critical By Default
- missing
<title>; - missing meta description;
- missing or duplicate canonical;
- no JSON-LD on pages expected to have schema;
- invalid JSON-LD;
- NewsArticle without usable image;
- public page blocked or unreachable during live check;
- report claims success without fetching or reading target.
Warning By Default
- missing hreflang when a multilingual equivalent is expected;
- missing OG/Twitter social tags;
- missing RSS autodiscovery;
- missing viewport;
- article meta tags missing when an article schema exists;
- no visible breadcrumb when BreadcrumbList is expected;
- performance metrics not measured.
Safety And Privacy Boundaries
- Do not mutate target sites.
- Do not use credentials unless the user explicitly provides them for this task.
- Do not store cookies, tokens, raw private analytics, raw IP logs, or private Search Console rows in reports.
- Do not claim ranking, indexing, citation, or crawler behavior from one HTML check.
Validation
Validate skill edits with:
bash
python3 $CODEX_HOME/skills/senior-skill-architect/scripts/lint_production_skill.py $HOME/SKILL/skills/seo-regression-validator
python3 $HOME/SKILL/skills/seo-regression-validator/scripts/audit_static_seo.py $HOME/SKILL/skills/seo-regression-validator/fixtures/good-news.html
python3 $HOME/SKILL/plans/seo-llm-skill-cluster/scripts/lint_skill_cluster.py $HOME/SKILLForward tests live in evals.json.
Output Shape
Return:
- Targets checked.
- Critical findings.
- Warnings.
- Skipped checks.
- Evidence paths or URLs.
- Implementation owner for each fix.
