Passion Project
A personal AI platform orchestrating 10 specialized agents, 264+ tools, and autonomous operations to manage an entire homelab infrastructure — built from the ground up as a creative engineering challenge.
An AI-Powered Homelab Management System
Thor is a comprehensive AI platform built from the ground up to manage, monitor, and automate an entire homelab infrastructure. Powered by Claude AI and the Model Context Protocol (MCP), Thor orchestrates 10 specialized AI agents — each with distinct personalities, capabilities, and areas of expertise — working together as a coordinated team to handle everything from security monitoring to media management.
What started as a personal automation project evolved into a full-scale multi-agent AI system with 264+ tools, multi-platform access, persistent memory, and autonomous operation capabilities. It represents the convergence of AI engineering, infrastructure expertise, and creative problem-solving.
Each agent has a distinct personality, specialized tools, and a defined area of responsibility.
The central brain — coordinates all agents, handles most tasks directly, and has access to the full suite of 264+ tools across 30 groups.
The guardian — monitors threats via SIEM, IPS, and NVR systems, providing real-time security awareness and autonomous incident response.
The engineer — manages virtual machines, containers, Kubernetes clusters, networking, and all infrastructure components.
The homekeeper — controls smart home devices, automations, lighting, and environmental systems through Home Assistant integration.
The strategist — a dedicated assistant for business operations including website management, content strategy, and digital presence.
The curator — manages the entire media pipeline from content discovery through availability on the media server.
The explorer — a research-focused agent for long-term planning, relocation research, and multi-factor decision support.
The advocate — maintains detailed care documentation, tracks appointments, and supports family health advocacy efforts.
The craftsman — handles professional task tracking, development work, and productivity optimization.
The adversary — conducts penetration testing, OSINT gathering, and red team operations from a dedicated security environment.
Built on a modern, extensible architecture designed for reliability and scalability.
Foundation language models powering all agent reasoning and decision-making, with both Opus and Sonnet tiers for cost-performance optimization.
Standardized protocol enabling AI agents to interact with external tools and services through a unified interface with 264+ tool implementations.
Monorepo codebase with shared core library, MCP server, Discord/Telegram bots, and Hono web dashboard — all type-safe with Zod validation.
Persistent context with daily logs, session management, semantic search, vector embeddings, and cross-agent bulletin boards for shared awareness.
Autonomous monitoring, proactive alerting, and cron-based task execution — agents operate independently without human prompting.
Agents create branches, write commits, and open pull requests autonomously — maintaining a complete audit trail of all changes.
Discord / Telegram / Web / CLI
|
[ Thor Overseer ]
/ | | \
/ | | \
Heimdall Jarvis Ada Apollo ... (10 agents)
\ | | /
\ | | /
[ MCP Tool Server ]
264+ tools / 30 groups
|
Proxmox | K3s | Home Assistant | Wazuh
UniFi | Synology | Grafana | GitHub | ...
Interact with the system from anywhere — each platform offers the full range of capabilities.
Slash commands for team interaction, notifications, and collaborative operations.
Mobile-first access for on-the-go infrastructure management and alerts.
HTMX-powered dashboard with real-time agent status and system overview.
Direct terminal interface — the primary power-user interface for all operations.
This is not a demo — these are real accomplishments from a system running in production every day.
Detects and responds to security threats autonomously — from SIEM alert triage to automated firewall rule creation and incident investigation.
Manages a 9-node Kubernetes cluster running approximately 190 pods via ArgoCD GitOps, with full lifecycle management and monitoring.
Found and fixed bugs in its own codebase autonomously, including a SIEM feedback loop that was generating false alerts and network routing issues.
Meal planning for dietary restrictions (Celiac disease), appointment coordination, and comprehensive care documentation for advocacy.
Manages an entire media pipeline from content search through library availability — fully automated from request to playback readiness.
Built Grafana dashboards, Prometheus alert rules, and complete monitoring stacks autonomously — from design to deployment.
The full technology landscape powering the Thor platform.
From a simple automation script to a production-grade multi-agent AI system.
Thor has been in continuous development, growing organically as new challenges presented themselves. Each phase built on the previous one, with the agents themselves contributing to their own evolution.
Initial concept — exploring Claude AI for homelab automation. Built first MCP tools for Proxmox and Docker management.
Expanded to multiple specialized agents. Added Heimdall for security, Jarvis for infrastructure, and Ada for home automation.
Launched Discord bot, Telegram bot, and web dashboard. Implemented persistent memory system and cross-agent communication.
Agents began operating autonomously — scheduled monitoring, proactive alerting, and self-healing capabilities.
Consolidated to 10 agents with clear specializations. Reached 264+ tools across 30 groups. Full infrastructure coverage.
Production-grade system managing all aspects of a complex homelab — from security to media to family support.
This project demonstrates AI engineering, infrastructure automation, and creative problem-solving at scale.
Get in Touch