Skip to content

Releases: foxsmith68-stack/AlexeiAI

AlexeiAI Foundation Release v0.1

30 Apr 13:47
0fec62e

Choose a tag to compare

Alexei v0.1 — Foundation Release Notes

Overview

Alexei v0.1 is the first public release of a local, persona‑driven AI agent written in C++ and powered by llama.cpp.
This release establishes the core architecture required for future learning capabilities — including persistent memory, persona definition, modular components, and a stable inference pipeline.

Alexei runs entirely on your machine with no cloud dependencies.

What’s Included

Core Architecture

Modular C++ agent framework

Clean inference loop built on llama.cpp

Configurable sampling and prompt pipeline

HTTP interface for message/response handling

Persona System

persona/identity.txt defining Alexei’s character

Configurable persona parameters

Style and behavior scaffolding

Memory System

Structured memory format (facts, preferences, traits)

Runtime memory storage directory

Template memory.json for initialization

Project Structure

Organized directories for src/, config/, persona/, memory/, models/, and third_party/

CMake build system

MIT license

Initial README

What’s Not Included Yet

This release intentionally focuses on the foundation.
The following features are planned but not implemented:

Automatic memory extraction

Adaptive learning from conversation

Persona evolution

Relevance scoring and memory pruning

Tool execution and autonomous behaviors

These will begin rolling out in v0.2 and beyond.

Why Release Now

This version provides:

a clean, stable base

a clear architecture

a working agent

a defined memory and persona system

a solid starting point for contributors

It’s the right moment to open the project to the community while keeping the roadmap forward‑focused.

Roadmap Highlights

v0.2 — Learning Begins

Fact and preference extraction

Automatic memory updates

Basic persona adaptation

v0.3 — Behavioral Evolution

Trait weighting

Style evolution

Memory relevance scoring

v0.4 — Tools & Autonomy

Tool execution

Multi‑step reasoning

Expanded HTTP API

Known Limitations

No automatic learning yet

Memory is static unless manually edited

Persona does not evolve dynamically

Requires manual placement of GGUF model