morphiq-build

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Pipeline Position

流程位置

Step 3 of 4 — consumes morphiq-rank output.

Input: Prioritized Roadmap (JSON) OR user prompt OR existing content.
Output: Build Output (JSON + artifacts) → consumed by morphiq-track.
Data contract: See
```
PIPELINE.md
```
§3 for the Build Output schema.

4步流程中的第3步——接收morphiq-rank的输出。

输入： 优先级路线图（JSON）或用户提示或现有内容
输出： 构建输出（JSON + 产物）→ 供morphiq-track使用
数据契约： 构建输出的schema请参考
```
PIPELINE.md
```
第3节

Purpose

用途

Morphiq Build fixes issues from morphiq-rank. It creates new content, optimizes existing content, generates schema markup, builds policy files, and produces artifacts to improve AI visibility. The core engine is a 6-step content lab pipeline.

Morphiq Build修复morphiq-rank识别出的问题，可创建新内容、优化现有内容、生成schema标记、构建策略文件、产出各类产物以提升AI可见性，核心引擎是6步内容实验室流程。

Entry Points

入口

Path A — From Prioritized Roadmap: Process issues by tier and priority. Route each to the appropriate fix workflow.

Path B — From User Prompt: Accept topic, optional source URLs (up to 5), optional ICP/brand context. Route to content lab pipeline.

Path C — From Existing Content: Accept content URL or raw text. Route to quality rewrite workflow.

路径A——来自优先级路线图： 按层级和优先级处理问题，将每个问题分配到对应的修复工作流。

路径B——来自用户提示： 接收主题、可选的源URL（最多5个）、可选的ICP/品牌上下文，分发到内容实验室流程。

路径C——来自现有内容： 接收内容URL或原始文本，分发到质量重写工作流。

Content Lab Pipeline (6 Steps)

内容实验室流程（6步）

Step 1: Ingest Sources

步骤1：源数据导入

Validate URLs, filter blocked domains, deduplicate, cap at 10. Accept raw text or PDF alternatives. Halt if zero valid sources.

验证URL，过滤被拦截域名，去重，最多保留10个。支持接收原始文本或PDF作为替代输入。如无有效源数据则终止流程。

Step 2: Extract Content

步骤2：内容提取

Crawl each URL → clean markdown. Extract title, content, outbound links, publish date. Halt if zero successful extractions.

爬取每个URL → 清洗为markdown格式。提取标题、内容、出站链接、发布日期。如无成功提取的内容则终止流程。

Step 3: Analyze Gaps

步骤3：缺口分析

Analyze against query space. Identify 5 gap types:

Gap Type	What Is Missing
Content	Unanswered questions, missing perspectives
Data	Missing statistics, quantitative evidence
Format	Wrong format for LLM retrieval
Depth	Surface-level, no expert insight
Fanout coverage	Sub-queries AI would chain but site cannot answer

Detect comparative intent. Evaluate fanout coverage using content type → sub-query rules. Generate up to 5 search queries.

For gap taxonomy and severity, read

references/gap-taxonomy.md

对照查询空间进行分析，识别5类缺口：

缺口类型	缺失内容
内容	未解答的问题、缺失的视角
数据	缺失的统计数据、量化证据
格式	不符合LLM检索要求的格式
深度	内容流于表面，无专业见解
辐射覆盖度	AI会链式查询但站点无法回答的子查询

检测对比意图，使用内容类型→子查询规则评估辐射覆盖度，最多生成5个搜索查询。

缺口分类和严重程度请参考

references/gap-taxonomy.md

。

Step 4: Research to Fill Gaps

步骤4：调研补全缺口

Run up to 5 live web searches. Collect authoritative sources, statistics (number + source + URL), expert quotes (speaker + credential), industry insights. If comparative intent, dedicate 1 search to brand data.

For citation rules, read

references/enrichment-sources.md

最多运行5次实时网页搜索，收集权威来源、统计数据（数值+来源+URL）、专家观点（发言人+资质）、行业洞察。如果存在对比意图，专门分配1次搜索查询品牌数据。

引用规则请参考

references/enrichment-sources.md

。

Step 5: Generate / Rewrite

步骤5：生成/重写内容

Produce final content applying Morphiq standard:

E-E-A-T signals, name-drop citations, expert quotes
Heading hierarchy, 50–75 word paragraphs, direct-answer blocks
Brand positioning (comparative or authority mode)
1,200–1,600 words, 5–7 H2 sections, FAQ with 3–5 Q&As
Minimum 3 statistics, 1 expert quote, sources section
No fabricated case studies

For full pipeline spec, read

references/content-lab-pipeline.md

遵循Morphiq标准产出最终内容：

E-E-A-T信号、明确标注引用来源、专家观点
标题层级结构、50-75词段落、直接回答区块
品牌定位（对比模式或权威模式）
1200-1600词、5-7个H2章节、包含3-5个问答的FAQ模块
至少3个统计数据、1条专家观点、来源板块
禁止虚构案例研究

完整流程规范请参考

references/content-lab-pipeline.md

。

Step 6: Validate Fanout Coverage (Fanout Issues Only)

步骤6：验证辐射覆盖度（仅辐射类问题）

For

fanout-*

issues with

fanout_context

: validate generated content addresses all triggering sub-queries and meets the competitive quality floor. If coverage < 80% or quality floor not met, revise once. Skip for non-fanout content.

Run

scripts/validate-coverage.py

with generated content + triggering sub-queries + quality floor.

对于带有

fanout_context

的

fanout-*

类问题：验证生成的内容覆盖所有触发的子查询，且达到质量竞争底线。如果覆盖率<80%或未达质量底线，执行1次修订。非辐射类内容跳过本步骤。

运行

scripts/validate-coverage.py

，传入生成的内容、触发的子查询、质量底线作为参数。

Post-Pipeline Processing

流程后处理

Process	What It Does	Reference
Schema Injection	Classify content type, generate JSON-LD. New content: embed schema in content artifact. Existing content: separate schema artifact with implementation tracking.	`references/schema-templates.md`
Metadata Optimization	Meta description, slug, OG tags	`references/metadata-patterns.md`
llms.txt Generation	Full autonomous pipeline: scrape → LLM → validate → repair → template fallback	`references/llms-txt-spec.md`
Content Restructuring	Fix headings, split paragraphs	—
Internal Linking	Link related pages for `site:` coverage	`references/content-lab-pipeline.md`
Enrichment	Additional search for missing stats/citations	`references/enrichment-sources.md`
FAQ Generation	Generate FAQ from gap analysis	`references/faq-guidelines.md`

处理环节	功能	参考文档
Schema注入	分类内容类型，生成JSON-LD。新内容：将schema嵌入内容产物；现有内容：输出独立的schema产物并跟踪落地情况	`references/schema-templates.md`
元数据优化	元描述、slug、OG标签	`references/metadata-patterns.md`
llms.txt生成	全自动化流程：爬取 → LLM处理 → 验证 → 修复 → 模板兜底	`references/llms-txt-spec.md`
内容重构	修复标题问题、拆分段落	—
内部链接	关联相关页面提升 `site:` 查询覆盖度	`references/content-lab-pipeline.md`
内容丰富	补充搜索缺失的统计数据/引用来源	`references/enrichment-sources.md`
FAQ生成	基于缺口分析生成FAQ	`references/faq-guidelines.md`

Issue Type → Fix Routing

问题类型→修复路径映射

Issue Category	Fix Approach
`agentic-*` schema	Schema Injection — generate JSON-LD
`agentic-*` metadata	Metadata Optimization — generate tags
`content-*` quality	Quality Rewrite — Step 5 pipeline
`chunking-buried-answer`	Quality Rewrite — Step 5 pipeline (Claude-driven rewrite to answer-first structure)
`chunking-*` structure (other)	Content Restructuring
`policy-*` files	Policy file generation
`fanout-*` coverage	Full 6-step pipeline for new content. When `fanout_context` is present, pass `triggering_sub_queries` to Step 3 and `competitor_sources` to Step 4. Run Step 6 (coverage validation) before post-pipeline processing.
`visibility-*`	Enrich existing content via pipeline

问题分类	修复方案
`agentic-*` schema	Schema注入——生成JSON-LD
`agentic-*` 元数据	元数据优化——生成标签
`content-*` 质量	质量重写——流程第5步
`chunking-buried-answer`	质量重写——流程第5步（Claude驱动重写为答案前置结构）
其他 `chunking-*` 结构问题	内容重构
`policy-*` 文件	策略文件生成
`fanout-*` 覆盖度	完整6步流程生成新内容。如存在 `fanout_context` ，将 `triggering_sub_queries` 传入第3步、 `competitor_sources` 传入第4步。流程后处理前运行第6步（覆盖度验证）
`visibility-*`	通过流程丰富现有内容

Build Output

构建输出

Artifacts with

type

: "content", "schema", "metadata", "policy_file". Each includes placement instructions.

产物带有

type

字段："content"、"schema"、"metadata"、"policy_file"，每个产物都包含落位指引。

Reference Files

参考文件

File	Purpose
`references/content-lab-pipeline.md`	Full 6-step pipeline with I/O formats
`references/gap-taxonomy.md`	Gap types, severity, search query rules
`references/enrichment-sources.md`	Citation format, source preferences
`references/schema-templates.md`	JSON-LD templates, skip conditions
`references/metadata-patterns.md`	SEO metadata rules
`references/llms-txt-spec.md`	llms.txt spec and generation
`references/faq-guidelines.md`	FAQ generation rules

文件	用途
`references/content-lab-pipeline.md`	完整6步流程及输入输出格式
`references/gap-taxonomy.md`	缺口类型、严重程度、搜索查询规则
`references/enrichment-sources.md`	引用格式、来源偏好
`references/schema-templates.md`	JSON-LD模板、跳过条件
`references/metadata-patterns.md`	SEO元数据规则
`references/llms-txt-spec.md`	llms.txt规范及生成规则
`references/faq-guidelines.md`	FAQ生成规则