✎ Edit this page

News

Announcements from BuddhistAI: new best skills and leaders, new benchmarks, releases, and new contributors and supporters. Newest first.

Follow updates on X: @buddhist_ai. You can also watch the BuddhistAI repos on GitHub or the Discussions. Maintainers: how to post is in the handbook.

17 November 2026 · Provenance for every score, tutorials and a governance page

Every score of the Tibetan table-of-contents skill now has a provenance record: the exact skill files, model, settings, prompt, data version, code, logs, tokens and cost. The standard applies to all new results. New in this knowledge base: four hands-on tutorials, a Governance page, a FAQ and a Glossary. The Directory now marks which datasets are open and which accept contributions. Organizations can now offer API credits or compute, and experts can review a benchmark.

17 November 2026 · Languages pages

The knowledge base now has a page for each of 21 Buddhist languages: the data that exists, who builds AI for it, and how well AI translates into it, with an overview of Buddhist countries. See Languages.

17 November 2026 · Directory of Buddhist AI projects and datasets

A Directory of Buddhist AI projects and datasets from all traditions and languages. Add yours by pasting a link.

13 November 2026 · First best skill: Tibetan table-of-contents extraction

The first task leaderboard went live: Tibetan table-of-contents extraction, on an 85-item benchmark from two Tibetan commentaries. Five skill variants were measured on Claude Haiku 5.5 and Claude Sonnet 5.5. The sa-bcad method skill, adapted from a colleague’s outliner skill, was released as toc-extraction-tibetan. It is close to the GEPA-optimized variant, which is the Best value on Haiku.

✎ Edit this page · History · Discuss