SQLite-based reducer kernel for evidence-backed agent learning protocols. Version 0.1.7 is the initial experimental release targeting Node.js >=24. Reduces agent trajectories into compact SQLite databases for traceable learning and debugging. Differentiates from general-purpose reducers by focusing on auditability, deterministic replay, and schema-enforced evidence logs. Development is early-stage with likely breaking changes.
npm install lyo-kernelNo compatibility data collected yet for this library.
Verified import paths — ran on the pinned version, not inferred.
Initialize an SQLite kernel, record an evidence entry, and query recent thoughts.
Upgrade Node.js to v24 or later.
Use a single Kernel instance per process or implement external lock.
Use pull the latest and update calls if API changes.
Use 'limit: 0' for no limit or omit the limit field.
Use import { createKernel } from 'lyo-kernel'; instead of import createKernel from 'lyo-kernel';Add "type": "module" to package.json or use .mjs file extension, and set --experimental-require-module if needed.
Ensure only one Kernel instance writes to the same .db file at a time.