|
The Ooloi documentation has grown considerably. ADRs, guides, READMEs, newsletters, blog posts – there's a lot of it, and not all of it is easy to navigate. So I've built something to help.
The Ooloi Librarian is a small RAG system – Retrieval-Augmented Generation, for the uninitiated – that answers questions about Ooloi's architecture and design directly from that documentation. It retrieves the relevant chunks, passes them to Claude Haiku as context, and returns a grounded answer with source citations. If nothing relevant is found, it says so rather than making something up. This is enforced in code, not by prompting. It's live now in the Ask the Librarian section of the Documentation page. Try it. Ask about the accidental rendering algorithm, or why Ooloi uses STM, or what the timewalker is. The answers come from the docs, not from the model's general knowledge. The backend runs on AWS – API Gateway, Lambda, Bedrock – rate-limited and CORS-restricted to this domain. Roughly four cents per thousand queries, and nothing at all when idle.
2 Comments
Magnus Johansson
25/2/2026 17:58:55
I asked the librarian what Flow Mode is and got an exhaustive answer. Very good.
Reply
Peter Bengtson
26/2/2026 08:31:48
Yes, it’s rather pleasing, isn’t it? What surprised me slightly is how useful the Librarian has become for my own development work. The AIs can consult it directly, instead of repeatedly ingesting entire documents and burning tokens in the process. I’m looking into that more seriously now.
Reply
Leave a Reply. |
AuthorPeter Bengtson – SearchArchives
February 2026
Categories
All
|
|
|
Ooloi is an open-source desktop music notation system for musicians who need stable, precise engraving and the freedom to notate complex music without workarounds. Scores and parts are handled consistently, remain responsive at scale, and support collaborative work without semantic compromise. They are not tied to proprietary formats or licensing.
Ooloi is currently under development. No release date has been announced.
|
RSS Feed