MCP Servers · chrisryugj

kordoc

kordoc is a TypeScript CLI and MCP server that parses Korean office documents (HWP3/5, HWPX, HWPML, PDF, Office formats) into Markdown, with features for document comparison, form-filling, and round-trip conversion back to native formats while preserving formatting.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/chrisryugj/kordoc

1.4k

GitHub stars

261

Forks

TypeScript

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	chrisryugj/kordoc
Owner	chrisryugj
Primary language	TypeScript
License	MIT — OSI-approved
Stars	1.4k
Forks	261
Open issues	2
Latest release	v3.17.0 (2026-07-06)
Last updated	2026-07-08
Source	https://github.com/chrisryugj/kordoc

What kordoc is

TypeScript-based document parser supporting multiple Korean and international formats; provides Markdown extraction, layout-preserving SVG rendering via cached coordinates or pure TS reflow, table structure recovery, form automation, and lossless patch operations on binary HWP and HWPX via selective cell/paragraph replacement without touching unmodified content.

Quickstart

Get the kordoc source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/chrisryugj/kordoc.git2cd kordoc3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Korean Government & Corporate Document Automation

Large-scale processing of HWP/HWPX documents common in Korean public administration and corporate workflows. Extracts structured text for analysis, comparison, and reuse without manual copy-paste or re-formatting.

AI-Assisted Document Workflow & Form Filling

Integration with Claude, Cursor, and other AI clients via MCP to enable LLM-driven document reading, analysis, template population, and generation. Preserves original formatting while AI handles content logic.

Multi-Format Document Ingestion & Conversion

Ingest heterogeneous document sources (PDF, XLS, XLSX, DOCX, HWP variants) and normalize to Markdown for downstream NLP, RAG, or content management pipelines. Handles complex tables, images, and embedded metadata.

Implementation considerations

Node.js 18+ runtime dependency; validate environment and package manager (npm/yarn) before deployment.
MCP setup wizard detects and configures AI clients (Claude Desktop, Cursor, VS Code, etc.). Manual JSON editing fallback required if detection fails.
Document parsing latency varies by format and complexity (tables, images, embedded objects); profile on representative samples before SLA commitment.
Lossless patch operations assume well-formed source documents; malformed HWP/HWPX may fail silently or produce warnings requiring manual review.
SVG rendering and reflow engine are TS-based; rendering large multi-page documents or high-frequency requests may require worker process pooling or caching.

When to avoid it — and what to weigh

No Korean Document Processing Need — If your documents are primarily English, European, or non-Korean formats without HWP/HWPX requirements, general-purpose document libraries may be more efficient.
Real-Time Streaming or High-Volume Throughput — kordoc parses documents sequentially. For petabyte-scale batch processing or sub-100ms response targets, dedicated distributed parsing infrastructure and benchmarking are required.
Strict Air-Gapped or Offline-Only Deployment — Setup wizard and MCP integration assume network access for npm package installation and optional AI client configuration. Fully offline deployments require pre-staging all dependencies.
Binary Patch Stability in Production Without Validation — Lossless patch operations (`patchHwpx`, `patchHwp`) modify binary files. Requires change validation and rollback procedures; unsuitable for systems with no QA gates.

License & commercial use

MIT License. Permissive open-source license allowing unrestricted use, modification, and distribution in both open-source and proprietary projects, subject to license and copyright notice retention.

MIT is a permissive OSI license. Commercial use is allowed. However, verify compliance with any bundled dependencies (e.g., PDF parsing libs, reflow engine) for transitive license obligations. Internal use, SaaS, and product bundling are permitted under MIT terms.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Strong
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Strong
Assessment confidence	High

Security considerations

No audit, penetration test results, or security advisories disclosed in data. Parsing untrusted documents (especially malformed PDFs or HWP files) should be isolated or sandboxed; v3.8.3 mitigates PDF parsing timeout (144s→2s). Patch operations modify binary files; implement version control and approval workflows. MCP integration transmits document content to AI clients; review data residency and confidentiality policies if handling sensitive government or corporate documents.

Alternatives to consider

LibreOffice / UNO

Universal document conversion and manipulation; no Korean-specific optimizations for HWP formats or direct Markdown export. Requires headless instance, slower startup, larger resource footprint.

Pandoc

Lightweight format conversion (DOCX, PDF, markdown); does not support HWP/HWPX, limited table recovery, no form-filling or AI agent integration.

Apache Tika

Multi-format text extraction; generic table handling, no HWP/HWPX support, minimal formatting preservation, no Markdown or patching capabilities.

Software development agency

Build on kordoc with DEV.co software developers

kordoc eliminates manual document parsing, form-filling, and format conversion in Korean government and corporate workflows. Integrate with Claude, Cursor, or your pipeline via CLI or MCP. Start with `npx -y kordoc setup` for AI client setup, or `npx kordoc <file>` for CLI-only use.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

markdownify-mcp jshookmcp chrome-devtools-mcp

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI Application Development API Development Custom Software Talk to DEV.co

kordoc FAQ

Does kordoc require Hancom (한컴 오피스) to be installed?

No. kordoc is a pure TypeScript parser for HWP binary formats and HWPX (ZIP-based XML). It reads Hancom's structure but does not invoke or depend on the Hancom application.

Can I use kordoc in production to auto-fill government forms?

Yes, for well-structured templates. The form-filling preserves original formatting (font, size, alignment). Test on sample forms first; complex nested tables or hand-drawn fields may require manual review or `--dx`/`--dy` offset tuning.

What happens if I edit Markdown and convert it back to HWPX?

If you use `markdownToHwpx`, a new HWPX is generated with Markdown formatting. If you use `patchHwpx` on the original, only changed cells/paragraphs are replaced in-place, preserving all original styling and unmodified content untouched.

Is the SVG rendering output production-ready for web display?

SVG output preserves layout fidelity for Hancom-cached documents (v3.10+). Multi-page rendering is supported (v3.14+). Reflow rendering (v3.15+) works for uncached files but may have minor differences from Hancom. Validate on sample documents before deploying as a preview service.

Work with a software development agency

DEV.co helps companies turn open-source tools like kordoc into production software. Our software development services cover the full lifecycle — architecture, web development, integration, and maintenance — delivered by software developers and web developers who ship. Engage our software development agency to implement or customize it for your mcp servers stack.

Ready to Automate Korean Document Processing?

Talk to DEV.co Schedule a Consultation