What problem does it solve?
Translating large Markdown documentation often disrupts formatting and code blocks, creating manual toil. This skill delivers an industrial-grade automated workflow that preserves code blocks, inline code, and document structure while translating to zh-TW, with self-healing batch handling and model-aware batching to minimize cost.
Core Features & Use Cases
- Self-healing batching: processes paragraphs in large batches and automatically falls back to single-paragraph translation when a batch fails.
- Model-aware optimization: uses different batch sizes for mini models versus flagship models to balance cost and precision.
- Mechanical masking: preserves block and inline code markers during translation to prevent tag-swapping.
- Modular pipeline: engine-core, file-io, and orchestrator components for reliability and maintainability.
- Use case: translate project documentation from docs/ into zh-TW, keeping Mermaid diagrams and code blocks intact.
Quick Start
Run the main pipeline to translate docs/ into zh-TW using an OpenAI API key and a preferred model.