使い方
- インストールコマンドをコピーし、自分の環境へ導入する前にソースを確認します。
- 必要なツール、認証情報、対応環境を原文で確認します。
- 入力と成果物を指定し、出力を資料と照合します。
npx skills add https://github.com/sergekostenchuk/ui-ux-agent-skill-system --skill internal-link-graph-architect開始前の確認
- 導入にはモデル契約、API枠、ファイルへのアクセス許可は含まれません。
- スター数や導入数は活動の指標であり、独立した品質・安全性試験ではありません。
リポジトリの原文
見出し、例、コード、表、リンク、参照画像を含む原文を表示しています。
Internal Link Graph Architect
Use this skill after information-architecture-seo has defined page roles and canonical policy.
Read references/link-graph-rules.md before producing a full graph.
Owns
- intentional internal link graph;
- brief-to-longform and longform-to-brief links;
- breadcrumbs;
- topic, project, author, community, and source-trail links;
- anchor intent;
- orphan-page checks;
- overlink and duplicate-intent risk checks.
Does Not Own
- canonical and hreflang policy;
- final URL architecture;
- external backlink placement;
- hidden links;
- visual UI layout;
- schema implementation.
Workflow
- Read
url-map.yaml, page roles, canonical model, language groups, and content model. - Identify required structural links: breadcrumbs, brief-longform pairs, topic/project/author/entity links.
- Add source-trail links only when they are visible and useful to users.
- Assign
link_type,anchor_intent,user_reason,seo_reason, andrisk. - Flag orphan pages, circular links without purpose, irrelevant exact-match anchors, and duplicated story-body routes.
- Produce
internal-link-graph.yamlusing assets/internal-link-graph.template.yaml. - Hand off schema/head implications to
technical-seo-schema-engineerand rendered navigation checks toui-ux-llm-product-architect.
Link Rules
- Every link must have a user reason.
- A short brief should link to its one longform article when a longform exists.
- A longform article should link back to its brief when the brief remains a standalone news surface.
- Topic pages organize related pages with summaries; they do not duplicate story bodies.
- Author and project links support trust and entity understanding.
- Breadcrumbs clarify hierarchy; they are not keyword stuffing.
- Related links should be sparse and defensible.
- External authority placement is out of scope.
Safety And Privacy Boundaries
- Do not recommend hidden links, cloaked links, injected footers, irrelevant exact-match anchors, or link stuffing.
- Do not link public pages to admin, private API, private analytics, raw IP logs, staging URLs, secret files, or unpublished drafts.
- Do not create extra public story pages to increase link count.
- Do not weaken UX or accessibility for SEO anchors.
Validation
Before marking work complete:
- every link has source URL, target URL, type, anchor intent, user reason, SEO reason, and risk;
- every page expected to be discoverable has at least one logical incoming link;
- brief and longform pages are paired without creating extra story bodies;
- topic/project/author links have entity or navigation purpose;
- hidden or irrelevant link recommendations are absent.
Validate skill edits with:
bash
python3 $CODEX_HOME/skills/senior-skill-architect/scripts/lint_production_skill.py $HOME/SKILL/skills/internal-link-graph-architect
python3 $HOME/SKILL/plans/seo-llm-skill-cluster/scripts/lint_skill_cluster.py $HOME/SKILLForward tests live in evals.json.
Output Shape
Return:
- Link graph summary.
- Required structural links.
- Optional related links.
- Orphan/overlink risks.
- Handoff notes for technical SEO/schema and UX.
