How to Use md2wechat-skill in Obsidian with Claudian

Installation and setup paths are distribution evidence, not proof of end-to-end compatibility with any named host. Treat each runtime as a validation candidate until its install, discovery, conversion, preview, and—when applicable—draft flow passes an end-to-end smoke test.

If you already write inside Obsidian, the best md2wechat workflow is not to leave Obsidian at all.

That is where Claudian becomes useful.

The one rule that prevents most Claudian failures

For Claudian, the install order should still be:

1. install the md2wechat CLI
2. install the shared skill
3. check discovery
4. fix GUI PATH only if needed

People often reverse the order. That is usually what creates the “the skill exists but the command still cannot be found” problem.

The shortest validation path

brew install geekjourneyx/tap/md2wechat
npx skills add https://github.com/geekjourneyx/md2wechat-skill --skill md2wechat
md2wechat version --json
md2wechat capabilities --json

If you installed via the shell installer, remember:

export PATH="$HOME/.local/bin:$PATH"

Why Claudian is different from other Coding Agents

The skill format is shared with the Claude Code ecosystem, so the skill installation story is familiar.

But Claudian runs inside Obsidian, which means GUI environment rules now matter.

That creates one classic mismatch:

• the terminal can find md2wechat
• Claudian inside Obsidian cannot

That is not really a skill bug. It is usually a GUI PATH inheritance issue.

How to fix the PATH problem

In Claudian, open:

• Settings
• Environment
• Custom variables

Then add your CLI directory, for example:

PATH=/Users/your-name/.local/bin:existing-path

That is often the missing piece when the runtime works in Terminal but not inside Obsidian.

The best first task to try inside Obsidian

After setup, start with a preview request:

Please use md2wechat to convert the current Markdown note into WeChat HTML and show me a preview first.

If that works, then move up to stronger tasks:

• ask for AI mode output
• generate a cover prompt
• create a draft if credentials are already configured

What discovery still matters here

Even in Claudian, the runtime should still ask the CLI before it assumes anything:

md2wechat capabilities --json
md2wechat themes list --json
md2wechat providers list --json
md2wechat prompts list --kind image --json

This matters because Claudian is a note-taking environment, not a guarantee that the publishing runtime is configured correctly.

The mistakes to avoid

1. Installing the skill before the CLI

That creates the illusion of readiness while the runtime is still missing.

2. Debugging terminal output only

Claudian may still fail even when Terminal works, because the GUI PATH is different.

3. Jumping straight to draft creation

A plain preview task is the better first proof. Draft creation adds credentials and cover handling on top.

Where to continue

• Claudian
• md2wechat-skill Guide
• Coding Agents