AI Agent Hub
Back to plugins
🖥️

dsh-hwp

Client Updated 2026.09.15

Run the following command in DeepSeek Harness:

dsh plugin install kevin9327/dsh-hwp

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install kevin9327/dsh-hwp on your DeepSeek Harness host to install the plugin from https://github.com/kevin9327/dsh-hwp; no build step is needed, just restart dsh web afterwards.

About this plugin

HWP and HWPX documents are the de facto standard in Korean government offices, schools, and corporations, yet their binary or zipped-XML structure makes them invisible to any built-in file reader. dsh-hwp closes that gap by registering a single read_hwp tool that converts a document into Markdown or plain text the model can reason over: paragraphs, tabs, line breaks, tables rendered as Markdown, text boxes, captions, and footnotes or endnotes. The HWP 5.x compound-file reader, the HWPX ZIP reader, and the XML parser are small TypeScript modules built entirely on Node.js built-ins, so there are no native binaries and no runtime dependencies. Safety rails cover every axis — file size, decompressed size, record counts, nesting depth, and output character budget are all capped, and long documents are served in paginated windows via offset and limit. An optional external converter can be plugged in for legacy formats such as HWP 3.x; it is spawned directly without a shell, with only a temporary file path substituted into its arguments.

This plugin is for developers who build agents in DeepSeek Harness and need them to work with Korean-language documents. Whether you are extracting application eligibility criteria from a government notice, summarizing a corporate form, or pulling data out of a school circular, a single natural-language prompt is all the model needs once the reader is in place.

Use Cases

  • Extract application eligibility and deadlines from Korean government notices
  • Read corporate form tables and summarize key dates
  • Parse school circulars to extract enrollment requirements

Best For

  • Developers handling Korean office documents in DeepSeek Harness
  • Teams needing automated reading of Korean public-institution files
  • Agent builders creating multilingual document workflows