Azdaja

Mandatory for complete semantic classification, review, extraction, or reduction over a large file (over 1 MiB, over 200 records, or too large for one Read), and whenever the prompt names Azdaja, asks if it is installed or available, or names the az virtual-memory tool. Invoke before reading or solving natively.

kubet Updated
The same task run twice: without Azdaja, 205 steps and 39.4M tokens; with Azdaja, 71 steps and 5.7M tokens

File contents

kubet/azdaja commit 2d6f04385a

Frequently asked questions

npx get-skill add kubet/azdaja-2