Skip to content

README.en.md

Latest commit

 

History

History
209 lines (159 loc) · 8.39 KB

README.en.md

File metadata and controls

209 lines (159 loc) · 8.39 KB
Raw

IP Publisher

简体中文 | English

IP Publisher Logo

GitHub stars GitHub release MIT License Install from ClawHub

Turn product documentation, knowledge-base facts, SEO keywords, trend hooks, and an outline into audited articles + 7-platform publish packs before any publishing step.

👉 Install from ClawHub · Demo Request · Demo KB · Platform Matrix · Publish Package Contract

KB-first · Audit-gated · 7-platform bundle · Wechatsync-ready

IP Publisher English workflow motion demo

Why teams try it

Knowledge-base first
It starts from docs, FAQ, and release notes instead of an empty prompt.		 Audit before publish
Accuracy, keyword fit, structure, and platform rules are checked before packaging content.		 One topic, many channels
The same source topic is adapted into the canonical 7-platform bundle by default.		 Built for operators
It returns a draft, an audit report, a publish package, and draft-sync hints instead of raw copy alone.

Traditional workflow vs IP Publisher

Traditional workflow IP Publisher
Keywords, trends, and source material live in separate places KB docs, keywords, trends, and outline are normalized into one structured request
Generate first and hope manual review catches problems Audit gate comes before publish packaging
Every channel needs manual rewriting The canonical 7-platform bundle is generated by default
Output often looks like generic AI rewriting Technical content leans toward Q&A, tables, heading hierarchy, and entity labels
Operators receive copy only and must rebuild the rest Operators receive article.md, audit_report.json, publish_package.json, and platform files

30-second overview

You provide The workflow does You get back
Product/tool KB docs, primary keywords, trend hooks, outline brief, audience Generates from the KB, audits grounding/keyword fit/structure/platform rules, then adapts the same core piece into 7 platform payloads article.md, audit_report.json, publish_package.json, and platforms/*.md
Inputs overview Outputs overview
Inputs
KB docs, keywords, trend hooks, and outline brief		 Outputs
Draft, publish package, and platform files
Audit report overview 7 platform payload overview
Audit report
Grounding, keyword fit, structure, and platform gates		 7 platform payloads
One source topic adapted into the canonical bundle		 
outputs/<task_id>/
  article.md
  audit_report.json
  publish_package.json
  platforms/
    wechat_official.md
    xiaohongshu.md
    zhihu.md
    juejin.md
    csdn.md
    toutiao.md
    weibo.md

See a real before / after

Before and after IP Publisher workflow

This example comes directly from the bundled repo demo:

Why this is not just another AI rewriter

  • It starts from a knowledge base, not an empty prompt.
  • It runs an audit gate before packaging content for distribution.
  • It defaults to a canonical 7-platform bundle, not an inconsistent 3-platform vs. 29-platform story.
  • It is draft-sync first through Wechatsync, not a username/password autopublisher.
  • For technical content it explicitly optimizes for Q&A, comparison tables, clear heading hierarchy, entity labels, and reproducible code examples.

Who this is for

  • Product, SEO, and knowledge-base content teams
  • Operators who need one source topic adapted into WeChat, Xiaohongshu, Zhihu, Juejin, CSDN, Toutiao, and Weibo
  • Teams that want accuracy and auditability before adding publishing automation

Where it fits best

  • Turning product docs and KB content into AI-friendly and SEO-ready article systems
  • Combining a trend hook with product knowledge into a reviewable draft
  • Splitting one technical topic into WeChat, Zhihu, Juejin, CSDN, Xiaohongshu, Toutiao, and Weibo versions
  • Giving operations teammates a lower-friction workflow with audit gates instead of auto-post risk

What it does now

IP Publisher now centers on one production path:

  1. Retrieve facts from a product/tool knowledge base
  2. Combine them with SEO keywords, trend hooks, and an outline brief
  3. Generate an AI-friendly article structure
  4. Audit grounding, keyword coverage, structure quality, fact density, and platform limits
  5. Export 7 platform payloads plus a publish package and Wechatsync draft-sync metadata

The default canonical bundle is:

  • wechat_official
  • xiaohongshu
  • zhihu
  • juejin
  • csdn
  • toutiao
  • weibo

Quick start

git clone https://github.com/veeicwgy/ip-publisher.git
cd ip-publisher
bash scripts/setup.sh
python3 scripts/quickstart.py

The new quickstart asks for:

  • product/tool name
  • primary keywords
  • trend/topic hook
  • outline brief
  • target audience
  • content type (general or technical)

It no longer asks “which platform?” as the main question. One topic now defaults to the 7-platform bundle.

Useful demo inputs:

Outputs

After a run you will get:

  • request.json
  • draft.json
  • audit_report.json
  • publish_package.json
  • article.md
  • platforms/*.md

Direct publishing stance

This repo does not do username/password login automation.

The recommended publishing bridge is Wechatsync:

  • browser-login based
  • draft-first
  • same web APIs as the platform editors
  • only after audit_report.status == pass

Humanizer

The repo now includes a lightweight built-in humanizer step inspired by Humanizer-zh. It focuses on reducing template-like AI phrasing while preserving facts.

Docs

License

MIT