Skip to main content

Literature Translator

Purpose​

Translate a literature attachment (Markdown or PDF) into a specified target language, producing a translated Markdown file and bilingual alignment data. It is strongly recommended to convert PDFs to Markdown first using MinerU for significantly better translation quality.

Inputs​

ParameterRequiredDescription
target_languageNoTarget language (default: zh-CN). Supported: zh-CN, en-US, ja-JP, ko-KR, de-DE, fr-FR, es-ES, ru-RU. Custom values are accepted.
modeNoTranslation mode: fast (default) or high_quality (slower).

Accepted attachment types: text/markdown, text/x-markdown, text/plain, application/pdf.

Trigger Methods​

  • Select an attachment directly (PDF or Markdown)
  • Select a parent item — the plugin automatically finds the first qualifying attachment
  • Only one attachment per parent item is processed
  • Items with an existing translation in the target language are automatically skipped

Behavior​

  1. Resolve the target attachment from the selection (direct attachment or first qualifying attachment under the parent item).
  2. Check whether a translation artifact already exists for the target language; skip if so.
  3. Submit the translation job to the Skill-Runner backend with the literature-translator skill.
  4. The translation pipeline runs multiple stages: alignment analysis, translation execution, and QA verification.
  5. Write the translated Markdown file to the same directory as the source, named <source-name>_<target-language>.md, and create a linked attachment under the parent item.

The workflow is fully automatic and does not pause for user intervention.

Output And Apply​

ArtifactDescription
Translated MarkdownWritten alongside the source file as <source-name>_<target-language>.md
Linked attachmentCreated under the parent item pointing to the translated Markdown
Alignment dataBilingual alignment JSON in the skill workspace
GlossaryExtracted glossary JSON in the skill workspace
QA reportQuality assurance report JSON in the skill workspace

Estimated Duration​

File SizeEstimated Time
Short paper (≤10 pages)3–5 minutes
Standard (10–30 pages)5–10 minutes
Long paper (30+ pages)10–18 minutes

Model Recommendation​

A model with subagent delegation capability is recommended. The translation pipeline includes alignment analysis, translation execution, and QA verification stages. Subagent delegation enables parallel processing of these subtasks, significantly improving efficiency and translation consistency.

Dependencies​

  • Backend: Skill-Runner
  • Skill: literature-translator