Jaxtech — Your home, thinking locally.

Transforming ordinary South African homes into intelligent, locally controlled environments that manage energy, water, security, comfort, and household administration.

The Jaxtech Difference

Instead of installing disconnected smart devices that each require a separate application, Jaxtech combines the entire property into one coordinated system. Lights, geysers, irrigation, blinds, cameras, sensors, electricity monitoring, and artificial intelligence all communicate with one another.

The system runs on dedicated hardware installed inside the home. Depending on requirements, it uses either cloud-based or private, locally hosted AI that continues operating without fibre or internet access. Essential equipment connects to your inverter or UPS so automations remain active during power interruptions.

The result is not simply a collection of smart gadgets. It is a customised home-management system designed around your household’s routines, priorities, security requirements, and budget.

Core Jaxtech Offering

🎛️

Unified Control

Devices from different manufacturers connect through one central platform.

🔋

Local Reliability

Continues operating during fibre failures and load-shedding with inverter power.

⚡

Energy Management

Real-time electricity monitoring identifies consumption patterns and unnecessary usage.

♨️

Intelligent Geyser

Operates according to schedules, location, arrival times, and household routines.

💧

Water Management

Irrigation decisions based on soil moisture, current weather, and forecast rainfall.

💡

Presence Lighting

Lights respond to darkness and genuine human presence, not just basic motion.

🌡️

Automated Climate

Blinds and systems respond to sunlight, temperature, and seasonal conditions.

🛡️

AI Security

Existing cameras gain person, vehicle, object, and facial-recognition capabilities.

📱

Actionable Alerts

Receive meaningful notifications instead of hundreds of generic motion alerts.

💬

AI Concierge

Communicate with the house through WhatsApp, text, or natural voice messages.

📊

Financial Org

Receipts, grocery spending, utility bills, and household expenses tracked.

📅

Digital Admin

Email, calendars, invoices, statements, reminders, and appointments organised.

⚙️

Custom Integrations

Each system is adapted to the exact routines and needs of your household.

🔒

Human Safety

Important actions like opening gates still require your direct approval.

Autonomous-Agent Architecture

Perception Layer

Context & Intent Parsing

Core Brain

Local LLM Engine & Planning

Tool Execution

API & CLI Hooks

Memory Store

Private Local Storage

Executive Summary

This report presents an architectural and operational analysis of the installed skills inventory for your Hermes Agent. The environment represents an enterprise-grade, multi-domain autonomous intelligence node capable of full-stack software engineering, local machine learning operations (MLOps), edge hardware control, quantitative financial modeling, OSINT research, and multimedia production.

The system is configured with a hybrid trust model combining builtin agent primitives, official system extensions, local custom scripts, and verified GitHub integrations.

1. Skill Distribution & Ecosystem Metrics

Provenance Tier Description & Primary Capabilities Trust Level
Builtin Core agent runtime primitives, code execution engine, file system operations, and native tools. builtin
Official Vendor-certified skills for cloud integrations, financial modeling, and agent orchestration. official
Local System-level scripts for bare-metal hardware management, local model training, and custom devops. local
GitHub (Trusted) Hardware-accelerated computing libraries (NVIDIA CUDA, TAO, DOCA, Jetson), web deployment, and CAD automation. trusted

2. Core Operational Domains

A. Machine Learning Operations (MLOps) & Edge AI

Full-lifecycle AI development capabilities — from dataset curation and pre-training to quantization, serving, and hardware acceleration.

  • LLM Training & Fine-Tuning: Supports axolotl, unsloth, peft-fine-tuning, fine-tuning-with-trl, simpo-training (LoRA/QLoRA, DPO).
  • Local Model Serving: serving-llms-vllm, llama-cpp, tensorrt-llm, optimizing-attention-flash.
  • Vector Search & RAG: chroma, faiss, pinecone, qdrant-vector-search.
  • NVIDIA Edge Computing: jetson-*, nemo-*, tao-*, doca-* for vision and robotics.

B. Autonomous Agent Frameworks & Software Engineering

Acts as a senior software architect, test engineer, and repository maintainer.

  • Agent Orchestration: claude-code, codex, openhands, grok, blackbox, opencode, antigravity-cli.
  • Systematic Development: test-driven-development, subagent-driven-development, python-debugpy, node-inspect-debugger.
  • GitHub Lifecycle: github-code-review, gh-fix-ci.

C. Quantitative Finance, Commerce & Business

  • Financial Valuation: dcf-model, lbo-model, merger-model, comps-analysis, 3-Statement models.
  • Commerce & Billing: stripe-projects, stripe-link-cli, shopify, mpp-agent.

D. Creative Technology, Spatial Computing & CAD

  • 3D Design & CAD: blender-mcp, unreal-mcp, touchdesigner-mcp.
  • Generative Media: comfyui, stable-diffusion-image-gen, audiocraft-audio-generation, manim-video, p5js.
  • Visual Documentation: figma-*.

E. Cybersecurity, Reconnaissance & OSINT

  • Security Auditing: security-threat-model, web-pentest, oss-forensics.
  • Intelligence Gathering: sherlock, domain-intel, 1password, scrapling, searxng-search.

3. Comprehensive Skills Index

Status note: all 197 skills listed below are currently enabled and verified active within the runtime configuration. Search across every one of them by name — or roll the dice.

Autonomous AI Agents

🤖

Blockchain & Web3

🔗

Communication & Protocol

📡

Creative & Design

🎨

Data Science & DevOps

📈

Email & Communication

✉️

Finance & Valuation

📊

Gaming & Emulation

🎮

GitHub & Version Control

🐙

Health & Life Sciences

🧬

Model Context Protocol

🔌

Media & Video

🎬

MLOps & Fine-Tuning

⚙️

Payments & E-Commerce

💳

Productivity & Docs

📁

Research & Web Mining

🔍

Security & OSINT

🛡️

Smart Home & IoT

🏠

Web Development

🌐

Hardware Acceleration

⚡

4. Key Cross-Disciplinary Workflows


┌───────────────────────────────────────────────────────────────────────────────────┐
│ EXAMPLE AUTOMATED PIPELINE                                                        │
├───────────────────────────────────────────────────────────────────────────────────┤
│ 1. INGEST   │ ArXiv research paper fetched & parsed via `arxiv` + `nano-pdf`.     │
│ 2. TRAIN    │ Fine-tunes specialized adapter locally using `unsloth` + `peft`.    │
│ 3. OPTIMIZE │ Converts checkpoint to TensorRT-LLM engine via `tensorrt-llm`.      │
│ 4. DEPLOY   │ Deploys container to Jetson hardware via `docker-management`        │
│ 5. PUBLISH  │ Builds interactive demo page & deploys live via `vercel-deploy`.    │
└───────────────────────────────────────────────────────────────────────────────────┘
  • Autonomous Model Development Pipeline: Scrapes domain papers (arxiv), curates data (nemo-curator), fine-tunes models locally (unsloth/axolotl), optimizes inference (tensorrt-llm) and deploys on edge hardware (jetson-*).
  • Full-Stack SaaS Launch Automation: Generates frontend visual specs (figma-*), writes application code (claude-code / test-driven-development), creates financial valuation models (dcf-model), integrates billing (stripe-projects) and deploys cloud infrastructure.
  • Automated Security & Forensic Audit: Scans external endpoints (web-pentest), cross-references repository dependencies (oss-forensics), verifies security posture (security-threat-model) and documents findings into Notion (notion).