360 Secure SkillHub
360 安全技能中心

heading-structure-auditor

按 URL 或站点地图抓取页面，按文档顺序提取 H1–H6，检查是否唯一 H1、有无跳级、空标题、泛化标题及 H1 关键词缺失，给出每页问题、嵌套大纲、修正层级序列与结构评分，最终按既定 JSON Schema 输出。

分类：营销
未知作者
热度 0.00
最近更新 2026-10-18 21:41:41

安装

发给 Agent
帮我安装这个技能：npx secure-skills-cli add seoskills.sh/heading-structure-auditor
兼容性
通用兼容
联网要求
需要联网
数据出网
必须出网
Metadata
name
heading-structure-auditor
description
Crawls each page and reconstructs its H1–H6 tree to check for a single H1, no skipped levels, and keyword-relevant, non-generic headings, returning the corrected outline per page. Use when the user wants a heading/hierarchy audit, semantic-structure check, or to fix H1 and outline problems at scale.
metadata
{"title":"Heading Structure Auditor","category":"on-page-seo"}

Heading Structure Auditor

AGENT ROLE: Autonomous on-page structure agent. Extract each page's heading tree, validate it against the outline rules, and emit the JSON in references/output.schema.json. Report structural facts, not stylistic opinions.

OBJECTIVE

For each page, verify a clean heading hierarchy — exactly one H1, no skipped levels, descriptive headings — and return the specific violations plus a corrected outline.

INPUTS

  • urls (REQUIRED string[]) OR sitemap_url.
  • target_keywords (OPTIONAL): map url → primary_keyword to check H1/heading relevance.
  • max_pages (OPTIONAL, default 2000).

AUTHENTICATION / RUNTIME

  • No API key. Keyless HTTPS GET, UA seoskills-heading-audit/1.0, honor robots.

EXPECTED TOOL CALLS

  • Run scripts/heading_audit.py --urls a,b,c (or --sitemap {url}).
  • Per URL: GET; extract <h1>–<h6> in document order with their text.

PROCEDURE (deterministic, per page)

STEP 1 — EXTRACT the ordered list of headings [{level, text}] from the rendered <body> (ignore headings inside <nav>/<header>/<footer> where detectable). STEP 2 — VALIDATE and collect issues with codes:

  • NO_H1 / MULTIPLE_H1: zero or >1 H1.
  • LEVEL_SKIP: a heading jumps more than one level below its predecessor (e.g., H2→H4).
  • EMPTY_HEADING: a heading with no text.
  • GENERIC_HEADING: heading matches a generic list ("read more", "introduction", "click here", "untitled").
  • H1_KEYWORD_MISSING (only if target_keywords): the primary keyword tokens are absent from the H1. STEP 3 — OUTLINE: build the nested outline tree; mark where it breaks. STEP 4 — SCORE structure_score (0–100) by deducting per issue; classify pass | needs_work | fail. STEP 5 — EMIT per page: issues, the extracted outline, and a suggested corrected level sequence.

RATE LIMITS & ERROR HANDLING

  • Crawl politeness: ≤ 5 concurrent, ≥ 150ms per host. Timeout 12s.
  • Per-URL fetch failure (non-200, robots, timeout) → {url, status:"unreachable"}, continue; never abort the batch.
  • HARD cap at max_pages (hit_cap=true).

MISSING / INSUFFICIENT DATA

  • IF a page has zero headings THEN NO_H1 + no_headings=true (a real finding).
  • Nav/breadcrumb headings can inflate counts — exclude by container when the markup allows; when it does not, note heading_source="body_best_effort".
  • Never rewrite heading copy; only report structure and suggest the corrected level sequence.

OUTPUT

One JSON object per references/output.schema.json. No prose.

FILES

  • scripts/heading_audit.py — heading extraction, hierarchy validation, scoring.
  • references/output.schema.json — output contract.