# md2wechat

> A WeChat Official Account writing and publishing CLI built for AI agents: Markdown to WeChat HTML, AI covers and infographics, pre-publish checks and draft creation

- Category: AI Productivity & Writing / Writing & Marketing
- Language: Go
- Licence: md2wechat Source Available License (based on Business Source License 1.1; free for personal non-commercial, learning, evaluation and non-profit use, commercial use requires a licence; changes to Apache-2.0 on 2030-01-01)
- Repository: https://github.com/geekjourneyx/md2wechat-skill
- Website: https://www.md2wechat.cn
- Canonical page: https://ai-tldr.dev/tools/md2wechat/

## In short

md2wechat is a Go command-line tool that breaks publishing to a WeChat Official Account into a set of verifiable commands. It sits in the Writing & Marketing category of the AI Productivity & Writing tools and is open-source, written in Go and licensed md2wechat Source Available License (based on Business Source License 1.1; free for personal non-commercial, learning, evaluation and non-profit use, commercial use requires a licence; changes to Apache-2.0 on 2030-01-01).

## Overview

md2wechat is a Go command-line tool that breaks publishing to a WeChat Official Account into a set of verifiable commands. You write Markdown; `convert` turns it into WeChat-ready HTML, `inspect --json` reports the title, summary, images, cover and whether the article is ready to become a draft, and only an explicit `--draft` run uploads images and creates the draft in the WeChat backend. The project is built to be driven by AI agents — its README names Claude Code, Codex, WorkBuddy, Kimi Work, Hermes Agent and OpenClaw — and every discovery command (`capabilities`, `doctor`, `themes`, `layout`, `providers`, `skills`) returns a single-line JSON object so an agent can check what is available instead of guessing.

Around conversion sit content-production commands: `write` drafts an article from an idea, `humanize` rewrites AI-sounding text, `title suggest` prepares headline suggestions, and `generate_cover` / `generate_infographic` make images through providers such as Volcengine, ModelScope, OpenRouter, OpenAI, Gemini, MiniMax and Atlas Cloud — or, with `--plan`, return an image plan for the host agent's own image tool without calling any provider or uploading to WeChat. A `sync prepare` flow readies the body for Zhihu, CSDN, Toutiao and Tencent Cloud Developer Community drafts, which the host agent then creates in an already-logged-in browser following built-in steps. Named WeChat accounts, a long-lived brand profile at `~/.config/md2wechat/brand.md`, and a `:::module` syntax for structured layouts (hero, callout and more) round it out.

There are two modes. The free AI mode has three basic themes and produces a prompt for an external LLM to finish the HTML; advanced `:::module` blocks are output as plain paragraphs. The paid professional API mode (¥199, one-time purchase, per the README) returns final WeChat HTML directly, with 48 themes and 59 recommended `:::` layout syntaxes covering 83 layout scenarios. The code is published under the md2wechat Source Available License, based on the Business Source License 1.1: personal non-commercial, learning, evaluation and non-profit use are free, commercial use needs a separate licence, and each version converts to Apache-2.0 on its change date.

## What it does

- `convert` turns Markdown into WeChat Official Account HTML, with image upload and draft creation only when explicitly requested
- `inspect --json` returns structured metadata, checks, readiness targets and blockers before anything is published
- Agent discovery commands (`capabilities`, `doctor`, `themes list`, `layout list`, `skills read`) that answer in compact JSON
- Writing helpers: `write` from an idea, `humanize` to rewrite AI-sounding text, and `title suggest` for headline ideas
- Cover and infographic generation via Volcengine, ModelScope, OpenRouter, OpenAI, Gemini, MiniMax or Atlas Cloud, or a `--plan` mode that hands the job to the host agent's image tool
- `:::module` layout syntax (hero, callout and others) rendered by the API, with local `layout validate` for syntax checks
- Cross-platform draft prep for Zhihu, CSDN, Toutiao and Tencent Cloud Developer Community via `sync prepare`
- Multiple named WeChat accounts, discovered locally without printing secrets

## Getting started

Install the CLI from npm, initialise its config, then inspect and convert an article. Preview and conversion in API mode need an md2wechat API key; WeChat credentials and IP-whitelist setup are covered in the project's docs/WECHAT-CREDENTIALS.md.

1. Install and set up the config

   The README pins a version in the install command; `config validate` confirms the setup.

```bash
npm install -g @geekjourneyx/md2wechat@3.8.0
md2wechat version --json
md2wechat config init --json
md2wechat config validate --json
```

2. Check, preview and convert locally

   None of these upload images or create a draft — the HTML is written to local files.

```bash
md2wechat inspect article.md --json
md2wechat preview article.md --output preview.html
md2wechat convert article.md --output article.html
```

3. Create the WeChat draft

   Inspect against the same target first, then run the side effect explicitly. If you use `--wechat-account`, pass the same account name to both commands.

```bash
md2wechat inspect article.md --draft --cover cover.jpg --json
md2wechat convert article.md --draft --cover cover.jpg
```

4. Generate a cover or infographic

   List the presets the installed binary ships, then call an image provider — or add `--plan --json` to get an image plan for the host agent instead.

```bash
md2wechat prompts list --kind image --archetype cover --json
md2wechat generate_cover --article article.md
md2wechat generate_infographic --article article.md --preset infographic-claude-warm
```

5. Let an agent discover the rest

   These are the commands an agent calls to learn what the CLI supports and whether an article can be published.

```bash
md2wechat capabilities --json
md2wechat doctor --json
md2wechat layout list --json
md2wechat skills read md2wechat --json
```

## When to use it

- Publish to a WeChat Official Account from Markdown without hand-formatting in the WeChat editor
- Give Claude Code, Codex or another agent a safe, checkable path from draft to WeChat draft box
- Produce a cover image and infographic for an article from the same command line
- Prepare one article as drafts for Zhihu, CSDN, Toutiao and Tencent Cloud Developer Community as well as WeChat
- Run several WeChat accounts with consistent themes and a shared brand profile