iLang WeChat Writing Assistant

Problem

WeChat posts often fail in practical ways: weak openings, long paragraphs, vague adjectives, AI-like phrasing, missing tables, and formatting that does not match the mobile editor. Asking a model to “write a viral article” usually skips the hard parts—sourcing material, choosing an angle, and preserving the author’s voice. This skill reframes the task as a structural transformation: it works from supplied material rather than inventing content from scratch.

How It Works

The core is a set of 17 writing rules, including:
- opening: state the point immediately and include a number within three sentences
- data_first: prefer concrete numbers, and mark gaps with [📊]
- paragraph: keep paragraphs under three lines for mobile readability
- table_rule: include at least two real data-comparison tables per 1,000 words
- voice_marker: leave [💬] for the author to add tone words instead of forcing AI-sounding insertions

After drafting, it runs a 10-point self-check covering AI fingerprints, em-dashes, rhetorical questions, paragraph length, table density, URLs, and ending structure. It also generates an I-Lang-style cover prompt constrained to a 2.35:1 wide format, with no text, logo, brand name, or real faces.

Boundaries

The skill expects material and a point of view. It does not accept empty prompts, fabricate numbers, quotes, or personal experience, or insert URLs in the body. It also avoids asking for likes, shares, or saves. For reducing AI-like phrasing, pair it with in-session text editing, and use a front-end Markdown tool for final copying and formatting.