wewrite-review

Reviews and revises WeChat article drafts against task briefs, sources, and editorial criteria.

3.2k|513|Updated Mar 26, 2026
One-click install
npx skills add https://github.com/oaker-io/wewrite --skill wewrite-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: wewrite-review
Source: https://github.com/oaker-io/wewrite/tree/main/skills/wewrite-review
Command: npx skills add https://github.com/oaker-io/wewrite --skill wewrite-review

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Draft articles often misalign with the original brief, contain unverified facts or fabricated personal anecdotes, and lack editorial polish. This Skill performs a structured editorial review of WeChat article drafts, fact-checks claims against recorded sources, rewrites failing content directly, and only marks an article publishable after it passes the editorial gate.

Core Features & Use Cases

  • Task Alignment Check: Verifies the article answers the audience question, matches the thesis, and delivers the promised takeaway, deleting off-topic sections.
  • Fact and Source Verification: Cross-checks numbers, dates, quotes, and first-person anecdotes against claims.yaml and recorded sources, removing or narrowing unsupported statements.
  • Five-Dimension Editorial Scoring: Rates accuracy, viewpoint, usefulness, voice, and readability, then directly revises the draft (up to two rounds) instead of only giving suggestions.
  • SEO Metadata Generation: Produces one main title, two alternative titles, a 40-character digest, and 3-5 tags following WeChat SEO rules.
  • Use Case: After generating a draft with the WeWrite pipeline, run this review step to catch fabricated personal stories, fix unsupported claims, and receive a publishable final article with an editorial report.

Quick Start

Ask the agent to check this article and revise it until it passes the editorial review for publication.

Frequently Asked Questions about wewrite-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a WeChat article draft before publishing?

Run the review step on the draft, which checks task alignment, verifies facts against recorded sources, and scores five editorial dimensions. If issues are fixable, the draft is revised directly and re-reviewed until it passes.

How does the editorial review handle unverified facts or fabricated stories?

Every number, date, quote, and first-person anecdote is checked against claims.yaml and original sources. Unsupported content is deleted, narrowed, or rewritten as explicit inference rather than patched from model memory.

Can I use this to check an external article file?

Yes. When you provide an external file and only ask for a check, it produces a review report without modifying the article. Direct revision happens only for pipeline drafts or when you explicitly ask for optimization.

What happens if a draft fails the editorial review twice?

After two revision rounds, unreliable content is removed and claims are narrowed to produce a version that can pass. An article that does not pass is never marked as deliverable or publishable.

Does the review step generate images or handle article layout?

No. The review module explicitly does not trigger image generation or layout. It only produces the final text, titles, digest, tags, and an editorial report; visuals and formatting are separate optional steps.