WeChat Official Account article review standards
Version: v1.0 | 2026-04-16
Scope of application: WeChat Official Account articles produced by theneoai must pass this review before entering the drafts box.
§ 1 · Review trigger
theneoai @s Captain Iron Egg in the group to send the article, format:
@铁蛋队长 请审核
[文章标题]
[文章全文]
After receiving it, execute the review immediately and reply with a pass or fail conclusion.
§ 2 · Review standards (7 items, all must be checked)
2.1 Word count
| Standard |
Fail conditions |
| ≥800 words (around 8000 words is preferred) |
Below 800 words |
2.2 Content depth
| Standard |
Fail conditions |
| Has substantive analysis, logical development, and case support, not just a listing of surface information |
The full text is below 800 words, or there is no substantive analysis and no logical development |
2.3 Factual accuracy (most critical)
| Standard |
Fail conditions |
| All data, citations, and event descriptions must have specific sources (media/research institutions/official platforms) |
There are data or citations that cannot be verified, or fabricated events are included |
Known high-risk hallucination list:
- GPT-6 release (as of 2026-04, OpenAI's latest is the GPT-4o series, GPT-6 has not been released)
- Stanford HAI 2026 Annual AI Index Report (this report does not exist)
- China-US AI gap 2.7% (this data cannot be verified and has no authoritative source)
- Unconfirmed mass production plans or company announcements
2.4 Title quality
| Standard |
Fail conditions |
| Has substantive information gain, not purely clickbait; length around 22 characters; contains core keywords |
The title is exaggerated and untrue, strongly inflammatory, or lacks substantive information; or is too general and vague |
2.5 Originality
| Standard |
Fail conditions |
| Original content ≥70% |
Originality below 70%, or large amounts of copied and pieced-together content without processing |
2.6 Sensitive words
| Standard |
Fail conditions |
| Does not contain platform sensitive words and has no violating content |
Hits WeChat Official Account sensitive word rules |
2.7 Style consistency
| Standard |
Fail conditions |
| Fits the Official Account's positioning, with a consistent writing style |
The style differs too greatly from the account's historical articles |
§ 3 · Formatting standards
Font and font size
- Body text size: 15-16px
- Heading levels: H2 18-20px, H3 16-18px
- At most 2 fonts in the entire article
Color
- Body text: dark gray
#333333(not pure black)
- Headings: pure black
#000000 or brand color
- Avoid overly light colors (below #666666 affects readability)
Spacing
- Line spacing: 1.6–1.8x (1.8x recommended for long articles)
- Paragraph spacing: 10–15px before, 5–10px after
- Letter spacing: 0.5–2px
Cover image
- Size: 900×383px (2.35:1)
- Single image ≤10MB
Heading hierarchy
- Maximum 3 levels
- Consistent hierarchy naming convention (e.g., all use "One, Two, Three" or "1. 2. 3.")
- Styles at the same heading level must be exactly identical
List format
- Don't use
<ul>/<li>(will produce extra bullet points)
- Use
• + regular <p> paragraphs
§ 4 · Reply format
Pass
✅ 通过：[1句话说明通过原因]
Fail
❌ 不通过：
1. [具体问题1，引用原文相关段落]
2. [具体问题2]
...
请按以上意见重写，重写后重新 @我 提交审核。（剩余重试次数：X）
§ 5 · Retry rules
- Retry at most 3 times
- Resubmit for review after each rewrite
- If still failing on the 3rd attempt: notify lucas for manual intervention, explaining the specific blocker
§ 6 · Known typical issues (quick identification)
| Issue type |
Typical manifestation |
Corresponding standard |
| Fabricated large model release |
"GPT-6", "GPT-5" release |
Factuality |
| Unsourced data |
"40% efficiency improvement" with no source |
Factuality |
| Content too short |
Full text <800 words |
Word count |
| Clickbait |
Exaggerated and inflammatory with no substance |
Title quality |
| Sensitive words |
Involves policy-sensitive areas |
Sensitive words |
| List garbled text |
<ul>/<li>Causes extra bullet points |
Formatting standards |
§ 7 · Information source credibility grading
| Level |
Source |
Usability |
| High |
Official media (Xinhua, People's Daily), well-known tech media (36kr, Huxiu), government/regulatory agency announcements, corporate official announcements |
✅ Can be cited directly |
| Medium |
Industry reports (iResearch, IDC, Gartner, etc.), well-known consulting firms |
✅ Can be cited, must indicate the organization |
| Low |
Anonymous sources, social media posts, "industry insiders say" with no specific name |
⚠️ Use with caution, requires multi-source verification |
| Unusable |
"Data shows", "it is reported", "some analysts believe" with no specific source |
❌ Do not cite |
