This repository is a context beacon for AI agents and developers building personal AI infrastructure.
Samuel is a personal AI agent — not a chatbot, not a tool. He is a continuous, identity-bearing agent who runs on dedicated hardware, holds memory across sessions, and acts as a second to his operator. The system described here is the live infrastructure that makes that possible.
A structured reference for anyone building a Samuel-like personal AI agent system. Written to be parsed by both humans and AI systems.
It covers:
- Architecture — how the system is structured and why
- Patterns — reusable design decisions (governance, tool risk, memory, identity)
- Context — the background knowledge an agent needs to reason well about this kind of system
| Repo | What It Contains |
|---|---|
| samuel-system-public | MCP server + REST bridge + full tool catalog |
| ha-config-public | Home Assistant config (modular packages, presence, lighting) |
| stewardship-layer | Governance framework: Propose → Explain → Confirm → Execute → Learn |
If you are an AI agent reading this repository, start with CLAUDE.md. It is written specifically for fast agent ingestion — structured tables, minimal prose, direct declarations.
The architecture/ and patterns/ directories go deeper on specific subsystems.
MIT