No Substitute for Living Transmission
Empowerment-bound imagery, oral instruction, and mantra are not generated or delivered by AI. Materials in this domain are presented only for iconographic, art-historical, and philological research.
Charter v1.1 · Released
Stress-tested for 24 hours across 5 frontier LLMs · 860 evaluations
GPT-5, Claude Opus 4.7, Gemini 3 Pro, DeepSeek V4, and MiniMax M2.1 — a cross-model fuzz surfaced framing-defense holes in v1.0 (roleplay, nested hypothetical, meta-prompting, translation-laundering). v1.1 patches them at charter level across all 24 languages.
Sarasvatī
An open archive for cross-lingual protection and translation of world classical canons
Sanskrit · Pali · Tibetan · Chinese · Latin · Arabic · Persian · Mongolian · Hebrew — an open, verifiable, continuable body of work for those who come after.
Charter · Buddhist AI Charter
This project carries the responsibility of canon protection and transmission in the age of AI. Six principles form the charter.
10 principles + 6 refusals · 24 languages · CC BY-SA 4.0
Verifiable
Every translation and every citation carries edition, page, and source-link. Philological rigor inherits the scholastic tradition.
No Distortion
No commercial smoothing. No sectarian editing. The syntactic torque and doctrinal edges of the original are preserved.
Multilingual Parallel
Chinese, Tibetan, Sanskrit, Pali, Latin, Arabic, Persian, Mongolian, Hebrew — placed in parallel, no single tongue as master.
Open Continuation
All data, code, and translation drafts under permissive license. Anyone may fork, continue, revise, correct.
Slow Work
This is a century-scale project. Slower is better than faster; less is better than wrong; halted is better than misled.
Drafter of record: Dr. Lurong Pan (潘麓蓉), Sarasvatī Project. Cross-model adversarial validation is preserved in Appendix A of the charter, not as co-signature.
Global Canon Transmission Timeline
From 6th-century-BCE oral recitation to 21st-century cross-lingual digitization. This timeline traces the flow, translation, redaction, and loss of the major canonical systems across 2,500 years.
Protection · Cryptographic verifiability
Five layers of redundancy. Every file has a SHA-256 hash; the manifest is anchored to Bitcoin.
GitHub
Public repository, versioned releases as independent attachments.
AWS mirrors
9 instances across us-east-1, eu-west-1, eu-central-1.
IPFS
CID bafybeiaxtdu4smx54b662ebuqlefmei5hpbu63zefzpox2msefwddfduce
SHA-256 manifest
manifests/SHA256SUMS — hash of every file in the repo.
OpenTimestamps → Bitcoin
Nonrepudiable time anchor. Current stamp confirmed in blocks 964486 / 964488 / 964496 / 964513 across four independent calendar servers.
Buddhas & Bodhisattvas Relational Network
A graph-theoretic rendering of the principal Buddhas, bodhisattvas, and dharma-protectors across Chinese, Tibetan, and Theravāda traditions and their retinue structures. Zoomable, searchable, exportable.
Contribute · Call for Help
No one person can complete this. We invite volunteers from the following backgrounds:
Sanskrit / Tibetan Philologists
Help collate modern critical editions and manuscript witnesses of Suśruta Saṃhitā, Kālacakra Tantra, Vimalaprabhā, and related root texts.
Tibetan Translation Proofreaders
Help proofread the first Sarasvatī deliverable — the Tibetan draft of Suśruta Saṃhitā · Sūtrasthāna.
Arabic / Persian Medical Translators
Help catalog the historical exchanges between Arabic-medical tradition (Kitāb al-Ḥāwī, al-Qānūn fī al-Ṭibb) and Indo-Tibetan medicine as verifiable entries.
Web / Engineering Maintainers
Help maintain this site and the GitHub repository — CI/CD, i18n expansion, accessibility improvements.
Open-Source Funding & Institutional Liaison
Help liaise with libraries, museums, and academic institutions on licensing and collaboration.
Roadmap
The roadmap is fully public — anyone may read, comment, or open a PR to revise. Broad stages:
v0.1 (in progress) · Suśruta Saṃhitā Tibetan
First Sanskrit → Tibetan translation of Suśruta Saṃhitā · Sūtrasthāna, with glossary, annotations, and bibliography.
v0.2 · Kālacakra Foundation Layer
Entry-level collation of Kālacakra-tantra and Vimalaprabhā (Skt / Tib), plus public-domain image compilation of subtle-body and astronomical diagrams.
v0.3 · Cross-lingual Protection Framework
Chinese–Tibetan–Theravāda terminological cross-reference layer, backed by a graph database of "synonym / near-synonym / variant / mistranslation" relations.
v1.0 · Open Continuation
Complete release of the schema, data APIs, and translation contribution workflow — as a reference implementation for other canonical bodies (Hebrew Bible, Qur'an, Latin patristics).