Skip to content

Welcome

LAP-CHRIS\chris edited this page Jun 17, 2026 · 8 revisions

Krnl-AI Community

A cognitive runtime for building local-first agents with persistent memory, safety checks, evolving skills, developer tooling, and desktop P2P surfaces.

Krnl-AI Community is the open, local-first edition of Krnl-AI. It runs entirely on your machine without requiring hosted infrastructure. All state is stored locally via SQLite, and you can bring your own LLM provider (OpenAI, Ollama, Anthropic, etc.).

Quick Start

dotnet tool install -g KrnlAI.Cli
krnlai config set provider ollama
krnlai chat --local

What You Can Build

  • Local agents with your preferred LLM
  • Persistent memory backed by SQLite with semantic search
  • Skills that can be created, refined, exported, and installed
  • Safety-aware execution with 20 fundamental rules (R01-R20)
  • Predictive world models for simulation and planning
  • Causal reasoning for understanding cause-effect relationships
  • Continuous learning pipeline (memory → analysis → simulation → consolidation)
  • Editor integrations for VS Code and Visual Studio
  • Cross-platform desktop apps via WPF and Tauri
  • Peer-to-peer video sessions via WebRTC signaling

Packages

Package Purpose
KrnlAI.Cli Local CLI and interactive TUI
KrnlAI.Sidecar Local HTTP sidecar for tools and editors
KrnlAI.Contracts Public DTOs and interfaces
KrnlAI.Embedded In-process kernel for community apps

SDKs

SDK Language Source
Python SDK Python 3.10+ sdk/python/
.NET SDK netstandard2.0 sdk/dotnet/

Documentation Map

Section Description
Getting Started Install, configure, first run
CLI Reference All CLI commands and options
Architecture System design and principles
Cognitive Cycle 10-step cognitive processing pipeline
Memory System Episodic, semantic, procedural, working, emotional, autobiographical, and prospective memory
Safety System 20 fundamental rules, adversarial guard, ethics
Emotion System VAD model, pain/reward, emotional memory
SDK Guide Python and .NET SDK documentation
Sidecar API HTTP API endpoints reference and P2P signaling
Desktop Apps WPF and Tauri desktop applications, API keys, privacy, and P2P
Editor Extensions VS Code and Visual Studio extensions
Samples Example projects and patterns
Integrations LangChain, CrewAI, AutoGen, FastAPI
Contributing How to contribute to the project

Community

  • GitHub Issues — Bug reports and feature requests
  • GitHub Discussions — Q&A, ideas, and showcase (when enabled)
  • License — MIT

Clone this wiki locally