MagicMDArchive Chinese articles into Markdown packages

Convert WeChat Official Accounts, Juejin, CSDN, and public technical articles into polished Markdown packages with optimized media, cleaner code blocks, direct links, batch reports, and one shared core for the CLI, Agent Skill, and Python SDK.

Quick Start

Agent Skill

SDK Integration

Fast start

CLI, Agent Skill, or SDK: one conversion core

Start from a command line, a natural-language Agent request, or a Python backend. MagicMD keeps the output package consistent either way.

Instant conversion

uv tool install magicmd
magicmd "https://mp.weixin.qq.com/s/example"

Batch mode

magicmd batch urls.txt -o output/

Agent request

Use MagicMD to convert these public article links into Markdown packages, then tell me which ones need manual review.

Ready-to-use output

Output package

output/article-title/
├── article.md
├── metadata.json
├── extraction-report.json
└── images/
    ├── img_001.png
    └── img_002.png

Fast conversionBatch processingCustom outputAI knowledge input

CLIFor local archiving

Install once, convert one URL or a whole urls.txt list, and keep reports beside the Markdown.

Agent SkillFor natural-language batches

Ask Codex, Claude Code, or another agent to convert, inspect warnings, and flag manual review items.

Python SDKFor backend workflows

Import MagicMD from Python and store Markdown, metadata, reports, and media paths in your own system.

Deep WeChat optimizationDownloads images, extracts video links, filters editor assets, and keeps source metadata.

Direct, cleaner linksRestores redirect wrappers where possible, so Markdown links open the real target.

Polished code blocksRemoves copy controls, line numbers, language noise, and article chrome.

Traceable batch runsConvert a urls.txt file and inspect a report for failures, warnings, and review items.

Custom output rulesTune package folders, Markdown names, metadata, reports, and media paths.

CLI and Skill readyThe local CLI, npm wrapper, PyPI package, and Agent Skill share the same core.