Skip to content
Continuous Translation

Continuous Translation

Automated Translated Docs using GitHub Actions and Models
Getting Started Read the blog post

This website is translated using AI, check out the other languages via the dropdown in the upper right corner. You can also check out the translation dashboard to see the progress of GenAIScript.

AI-powered Translations

Section titled “AI-powered Translations”

GitHub Actions

When you commit changes to your documentation files, the translation workflow automatically starts a translation job.

GitHub Models

The AI workflow uses Markdown parsers and LLM inference to precisely and incrementally translate your documentation.

Next steps

Section titled “Next steps”

Add the Workflow

Add the translation workflow by telling your agent…

add the workflow for the pelikhan/action-continuous-translation action

Configure your site

Edit the workflow with the Astro Starlight information.

Translate!

Commit and let the workflow do the work!

Read the docs

Learn more in README.

How it works?

Section titled “How it works?”

The translation workflow uses a mix of Markdown parsers, AST manipulation and LLM inference to translate and validate your documentation files incrementally.

  1. 📄 Parse - Convert markdown to AST (Abstract Syntax Tree)
  2. 🔍 Analyze - Identify content that needs translation vs. existing translations
  3. 🤖 Translate - Use AI to generate high-quality translations
  4. ✅ Validate - Ensure translation quality and inject into document
  5. 💾 Cache - Save translations for future incremental updates
  6. 📝 Commit - Automatically commit changes to your repository

Features

Section titled “Features”
  • AI-powered translations for your documentation.
  • Seamless integration with GitHub Actions.
  • Incremental translation updates using LLM inference.
  • Uses Remark and AST manipulation for reliable translations.
  • Quality control with LLM-as-judge and compiler checks.
  • Supports Markdown, GFM, and MDX formats (to some extent).
  • Astro Starlight integration

This project is built using GenAIScript.