# Chinese Classics Library · 中华古籍文库 cn.jic.io

> Part of JIC. Index of all JIC sites: https://jic.io/llms.txt · 中文: https://cn.jic.io/llms-zh.txt · Agent handbook: https://jic.io/agents.txt

> A full-text library of public-domain Chinese classics, open to agents: the Daizhige ten canons (殆知阁十藏), 15,694 titles (about 2 billion characters, simplified Chinese);
> CBETA Chinese Buddhist texts, 4,882 works (traditional Chinese, with fascicle and page-column-line numbers); 344,270 poems through the dynasties.
> All public-domain content needs no login, no key and no quota request; every endpoint allows CORS from any origin and responses are cacheable.

## Two ways in

- HTTP JSON: `GET https://cn.jic.io/api` returns a self-describing list of endpoints; that one document is enough to use the whole library.
- MCP (Streamable HTTP, stateless): `https://cn.jic.io/mcp`.
  Claude Code: `claude mcp add --transport http jic https://cn.jic.io/mcp`
  Tools: search_fulltext, search_title, semantic_search, read_text, book_info, browse,
  resolve_cbeta, author, classify, poem, library_overview.

## Paths

Every book has one path: "canon/subfolder/title.txt", e.g. `史藏/正史/史记.txt`,
`CBETA/大正藏/般若部类/T0235 金刚般若波罗蜜经.txt`. The same path gives:

- Whole text: `/t/<path>` (text/plain, gzip, cached forever)
- Catalogue record: `/api/book/<path>`
- Text slice: `/api/text/<path>?offset=0&limit=20000` (ten canons) or `?juan=7` (CBETA)
- Reading page: `/book/<path without .txt>` (the page for people; link it when you quote)

Canons: 易藏 (Changes), 儒藏 (Confucian), 道藏 (Daoist), 佛藏 (Buddhist), 子藏 (Masters), 史藏 (Histories), 诗藏 (Poetry), 集藏 (Collections), 医藏 (Medicine), 艺藏 (Arts), CBETA.

## Common steps

1. Find a book: `/api/find?q=史记` → books[].path
2. Record and size: `/api/book/史藏/正史/史记` → chars, juans (CBETA), how_to_read
3. Read: `/api/text/史藏/正史/史记?offset=0&limit=20000` → text, next_offset;
   add `&format=txt` for plain text.
4. Find where a sentence comes from: `/api/search?q=元亨利贞&limit=10` → results[].snippets (pre/hit/post),
   hits, juan, lb (CBETA line number), text_url. candidates_total is the candidate count; page with offset.
   Narrow with `canon=史藏` or `canon=CBETA/大正藏`; `verify=0` returns the candidate list only.
5. Find Buddhist passages by meaning: `/api/sem?q=如何对治嗔恚&c=T`
6. Go straight to a sutra number: `/api/cbeta/T0235`, `/api/cbeta/T0262_007`, `/api/cbeta/T08n0235_p0749a05`
7. Authors and classification: `/api/people?q=朱熹`, `/api/people/朱熹`, `/api/class`, `/api/class/sibu/易类`
8. Poems: `/api/poem/12345`

## Notes on the texts

- The ten canons are simplified Chinese; CBETA is traditional. Full-text queries normalize simplified/traditional and variant characters; the texts themselves are not converted.
- Each CBETA fascicle starts with a line `#J<fascicle>`, and each line looks like `T08n0235_p0749a05║text`; `plain=1` strips the line-number prefix.
  Missing characters are marked `⟦|CB12345⟧`; the glyph image is at `/gaiji/CB12345.gif`.
- The ten canons were typed in by online volunteers over many years and have not been systematically collated; check an authoritative edition before quoting.

## Licence

- Ten canons: public-domain texts; data from daizhigev20 (https://github.com/garychowcmu/daizhigev20).
- CBETA: CC BY-NC-SA 3.0 TW — attribute CBETA, no commercial use, share derivatives under the same licence.
- Poems: chinese-poetry (MIT); the original texts are public domain.
- 《道祕天书》 is a contemporary work, licensed to JIC members only and not part of the open API.

## The library family

Every JIC library has the same interface: cn, jp, en, la, el, he, ar, fa, sa, pi, ko, de, fr, ru, es, it .jic.io,
and atlas.jic.io (World Language Atlas: parallel Bibles in 1,026 languages, Wiktionary etymologies, UD treebanks). Each has /llms.txt (English), /llms-zh.txt (Chinese), /api and /mcp.