Annotate classical Chinese texts word-by-word and render them into a long PNG image.
npx skills add https://github.com/lijigang/ljg-skills --skill ljg-classic
Please help me install this Agent Skill. Skill: ljg-classic Source: https://github.com/lijigang/ljg-skills/tree/main/skills/ljg-classic Command: npx skills add https://github.com/lijigang/ljg-skills --skill ljg-classic
Reading classical Chinese texts (文言文, 古诗文, 经史子集) requires juggling dictionaries, commentaries, and translations scattered across sources. This Skill produces a single long PNG image that combines the original text, per-token color-coded annotations, a chapter interpretation, and an optional thematic illustration, so the whole chapter can be read continuously in one artifact.
Ask the assistant to annotate a classical Chinese chapter word by word and render it as a long PNG, for example: run ljg-classic on 道德经第十六章 with full annotation and interpretation.
High-intent search queries and answers about installing and using this skill.
Provide the book name, chapter, and optionally the source text, then invoke the skill. It locks the base text, annotates every non-punctuation token, writes a chapter interpretation, and renders a verified long PNG to your Downloads folder.
Three semantic colors are used: teal for word-level meanings and grammar, vermilion for clause-level actions and causality, and ochre for textual variants, loan characters, and interpretation disputes. The chapter interpretation uses warm dark ink on a warm paper background.
Yes. The rendering and validation tools are TypeScript scripts executed with Bun, and screenshots are captured through Playwright's Chromium. First-time setup requires running bun install and bunx playwright install chromium in the skill directory.
Yes. Say you want a pure text version or no illustration, and the workflow omits the hero image fields. If image generation fails or is unavailable, the skill continues rendering a text-only version and notes the reason in the delivery report.
Validation fails when the reconstructed original text differs from the locked source, annotation coverage is below 100%, PNG dimensions mismatch the manifest, file hashes drift, or QA slices leave coverage gaps. All hard gates must pass before the PNG is copied to Downloads.
Do not use it for translating a single sentence or writing modern content; a plain translation or a general writing skill fits better. It is designed for full-chapter annotation and rendering, not quick one-line lookups.
SaberOnGoTranslate Italian books into Simplified Chinese EPUB editions with chapter-level QA.
tense-iTransform long-form documents into structured lectures with traceable evidence.
SelogazApply %%...%% instructions and rewrite ==...== passages in annotated text.
Clear-CapabilitiesExplain CVEs, GHSA findings, and scanner findings in plain English.
