MarkItDown

SkillFiles & storage

Uses the MarkItDown installed in the current environment to reliably convert files such as DOCX, PDF, PPTX, XLSX, images, HTML, and CSV into Markdown. Use when the user mentions "convert to md", "markitdown", "convert document to markdown", or "turn docx/pdf/pptx/xlsx into markdown".

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the MarkItDown skill

What this skill tells your AI

The instructions your AI receives, as published by seed-forge/harness-ai-kit in skills/markitdown/SKILL.md and read by ahel’s review.

这是当前项目的本地化 markitdown 技能入口。

它基于上游全局 Skill,但这里补上了本环境下已经验证过的稳定用法,避免每次重复处理安装和命令路径问题。

Use This Skill For

  • docxpdfpptxxlsx 转成 markdown
  • 批量把资料转换成更适合 LLM 处理的文本格式
  • 在 Obsidian 项目里落地文档转写结果

Environment Notes

  • 当前环境已验证 python -m markitdown 可用
  • 建议优先调用本地包装命令:
    • .\.agents\tools\markitdown.cmd
  • 如果直接输入 markitdown 失败,通常是因为 Python 的 Scripts 目录不在 PATH

Stable Invocation

.\.agents\tools\markitdown.cmd input.docx -o output.md

或:

python -m markitdown input.docx -o output.md

Verified Dependency Strategy

本环境里不建议直接依赖:

pip install "markitdown[all]"

因为实测可能把主包错误回退到旧版本。

更稳的方式是:

python -m pip install --upgrade markitdown==0.1.5
python -m pip install azure-ai-documentintelligence azure-identity openpyxl pandas pydub python-pptx speechrecognition youtube-transcript-api openai pathvalidate puremagic mammoth olefile xlrd pdfminer-six pdfplumber

Workflow

  1. 确认源文件存在,并避免覆盖现有 .md
  2. 优先输出到新文件名,例如 *.markitdown.md
  3. 转换后抽查文件头部,确认不是空文件
  4. 如需给 Obsidian 使用,再做一轮 markdown 清洗

推荐输出格式

执行完毕后按以下结构输出:

状态:✅ 成功 / ⚠️ 部分成功 / ❌ 失败

<输入→输出文件/格式/路径><值/状态>说明

下一步:<可执行动作>

Practical Notes

  • DOCX 已在当前环境验证通过
  • PDF、PPTX、XLSX、音频、YouTube 转写依赖已补齐
  • Azure Document Intelligence 和 OpenAI 增强能力仍需各自凭证

Signals

GitHub stars
22
Forks
2
Last commit
Aug 2026
Advanced
Catalog kind
skill
Gateway key
markitdown-seed-forge
Source
github.com/seed-forge/harness-ai-kit