Project: Automated Multi-language Technical Documentation Pipeline
(Made with DeepL MCP Server + DeepL API)
What I built
I created an automated translation pipeline that converts a large-scale Japanese technical documentation site into a fully usable English version using the DeepL API through the new DeepL MCP Server, integrated with a coding-agent workflow.
The Problem
I maintain a technical site with:
180 pages
1.5 million Japanese characters
Translating this manually would normally require:
A technical writer + translator
Several weeks (2β4 weeks)
A high cost (100Γ more expensive than DeepL API)
How it uses DeepL
Using the DeepL MCP Server, the agent:
Sends each section/page through the DeepL API
Collects and stores translations
Rebuilds the site automatically (VitePress)
Performs QA via round-trip checks
Outputs a full English version
Results
Total translated characters: 27,349
Translation time: ~1 day
Cost: 1,318 + 630 yen (~12 USD)
Translation quality: Extremely high, good enough for technical content
Final QA performed with DeepL + GPT/Claude
Why it matters
This workflow:
Makes multi-language documentation accessible for individual developers
Reduces translation costs dramatically
Eliminates the biggest barrier for solo creators who want to reach a global audience
Demonstrates the power of integrating DeepL MCP Server with coding agents
Links
Japanese site: https://shuji-bonji.github.io/RxJS-with-TypeScript/
English site (DeepL-translated): https://shuji-bonji.github.io/RxJS-with-TypeScript/en/
GitHub repo: https://github.com/shuji-bonji/RxJS-with-TypeScript