NOVA DEEP KNOWLEDGE Desktop Multi-Agent AI

by tomspeechnz in Circuits > Computers

246 Views, 2 Favorites, 0 Comments

NOVA DEEP KNOWLEDGE Desktop Multi-Agent AI

This is like having ChatGPT running on your Desktop of your Win11 PC!! It will do most things that big AIs can do (some the presentation is better, others it cannot match like creating diagrams). The point is to have your own AI on your PC which can then be used for electronic projects like running your lights of your house, videos,music or just doing your maths or computing homework! This is not a Chatbot however, it is a serious AI that can be used for serious research purposes. It will run with local Ollama models you can download or you can use powerful AI models from openrouter such as Claude. When using local Ollama models you have complete privacy and it will run off-grid but it will run slower and you will have less brain power than the online models. Though this all depends on the size of your graphics card on your PC. What it can't do at present is Video input but that isn't too difficult to add later.

Download Python 3.10 or Similar

Assuming you are new to this you probably don't have the Python programming language on your PC so you'd better first download it. Use version 3.10 or slightly above, don't take the latest version because latest isn't always best until all bugs are ironed out! Download directly from https://www.python.org/downloads/windows/ and get the 64 bit version as most modern PCs are 64 bit unless you know otherwise about your machines. IMPORTANT: Make sure as in the picture that you tick the Add python to path box. Otherwise you will avoid later pain! See the picture. THEN check that it is installed after a reboot by going into the command prompt i.e CMD and typing python. It should respond a bit like this depending on your PC paths:

C:\Users\OEM>python

Python 3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)] on win32

Type "help", "copyright", "credits" or "license" for more information.

>>>

If it doesn't then you haven't got the path installed and the rest won't work. You will have to manually install the path. Also don't be tempted to download the Windows version of Python from the Windows store as that won't work on this project either.

Download Git Program

Go to the link https://git-scm.com/install/ and download Git (assuming you don't already have it for other projects). This will enable you to download the project files from git (ie an online repository where I stashed the code!)

Download Visual Studio Desktop Tools 2022

We need the Visual Studio Desktop tools. We don't need the whole Visual studio as in the image. The link is Tools for visual studio 2022 (not 2026 as in the image)

https://aka.ms/vs/17/release/vs_BuildTools.exe

Install the above link only, we DO NOT need the whole of Visual Studio. This is needed for python. The pip command in python downloads the libraries you need but it may not have the right binary version for your setup so it is used to compile the 'wheels' i.e. source. You could run without it but there is a chance it may need to do some compiling. This is all done automatically invisible to the user so best to avoid trouble later.

Download Free Version of Pycharm

Download Pycharm (the free version, we don't need the bought full version). Download from https://www.jetbrains.com/pycharm/download/?section=windows . This is the final preparation step, the rest is just installing the actual software.

Test PyCharm

Navigate to where I put the main software: tommoirnz/NOVAKnowledge_5 and have a quick look, we will use this link in Pycharm, assuming you have installed it as the previous step.

First check Pycharm by running it and selecting File--> new project. Tick create a welcome script and then click the blue create button and your test program PythonProject1 will appear. Pycharm creates a test program

# This is a sample Python script.

# Press Shift+F10 to execute it or replace it with your code.

# Press Double Shift to search everywhere for classes, files, tool windows, actions, and settings.

def print_hi(name):

# Use a breakpoint in the code line below to debug your script.

print(f'Hi, {name}') # Press Ctrl+F8 to toggle the breakpoint.

# Press the green button in the gutter to run the script.

if __name__ == '__main__':

print_hi('PyCharm')

# See PyCharm help at https://www.jetbrains.com/help/pycharm/

(you don't need to type this). Press the green Run button at the top and Hi, PyCharm should appear in the command window at the bottom.

C:\Users\OEM\PycharmProjects\PythonProject1\.venv\Scripts\python.exe C:\Users\OEM\PycharmProjects\PythonProject1\main.py

Hi, PyCharm

Process finished with exit code 0

You are now half way there. We now have to download the git. Go to File -->close project to get out of your test program.

Clone the AI Repository to Your PyCharm Directory

Once you have closed the previous test project you will see the above appear on PyCharm and the top right button which says Clone Repository must then be pressed.

Make Sure You Run in a Virtual Environment .venv

Type the URL in as shown tommoirnz/NOVAKnowledge_5 and press the blue clone button. All the files will appear but don't be tempted to run just yet. The reason is as follows. PyCharm can sort of run like that but it needs a virtual environment otherwise all sorts of bad things happen!! You know there is a virtual environment when a file right at the top says .venv. If it is not there then don't run. If it does appear then you will only need to setup the Python interpreter in the next step.

Select the Python Interpreter

On PyCharm select File-->settings--->Python then select the blue Interpreter link. When clicked another menu appears. Select Python 3.10 or the version you downloaded. Also check the .venv file is top left in the program listings of the code. If it's there you can run otherwise go back to File--> settings and go to this link for further instructions. Configure a virtualenv environment | PyCharm Documentation

This can be fiddly sometimes but once set up it is easy to proceed.

Run the Program MAIN_RUNME.py

from the list of files on the left of PyCharm select MAIN_RUNME.py and the program will run! However it won't run properly until you edit a few files.

Purchase Openrouter Key or Download Free Ollama Models

Go to OpenRouter and buy some time on your chosen AI. I use Claude 4.6. If you click on config.json you will see

{

"cache_directory": "C:\\code_cache",

"error_cache_file": "error_search_cache.json",

"cache_max_age_days": 300,

"cache_max_entries": 1000,

"default_model": "anthropic/claude-sonnet-4-6",

"max_tokens":32000,

"cloud_models": {

"provider": "openrouter",

"api_key": "",

"base_url": "https://openrouter.ai/api/v1",

"models": [

{"display": "Qwen3-235B-A22B (Coder)", "id": "qwen/qwen3-235b-a22b"},

{"display": "Qwen2.5-Coder-72B", "id": "qwen/qwen2.5-coder-72b-instruct"},

{"display": "DeepSeek-R1", "id": "deepseek/deepseek-r1"},

{"display": "DeepSeek-V3", "id": "deepseek/deepseek-chat-v3-0324"},

{"display": "Qwen3.5-122B-A10B (paid)", "id": "qwen/qwen3.5-122b-a10b"},

{"display": "Qwen3-Coder-480B (paid)", "id": "qwen/qwen3-coder"},

{"display": "Qwen3-Coder-480B (free)", "id": "qwen/qwen3-coder:free"},

{"display": "Claude Sonnet 4.6", "id": "anthropic/claude-sonnet-4-6"},

{"display": "Claude Opus 4.6", "id": "anthropic/claude-opus-4-6"}

]

}

You see that is my default "Claude Opus 4.6", "id": "anthropic/claude-opus-4-6". You can select any one you like from the dropdown menu when NOVA runs. But unless you have installed Ollama on your pc and downloaded the models to disk (by the pull command) you will be running off the Cloud and openrouter as default. You will need a key from https://openrouter.ai which you must buy credits. (if you run local models of course it costs you nothing). For local models first install https://ollama.com/download for Windows. Then in PyCharm on the command window pull the model you want. For example Ollama pull <model-name>

Examples:

Llama 3.2

2.0GB

pull llama3.2

Llama 3.2

1.3GB

pull llama3.2:1b

Gemma 3

3.3GB

pull gemma3

DeepSeek-R1

4.7GB

pull deepseek-r1

Phi-4

14B

9.1GB

pull phi4

Mistral

4.1GB

pull mistral

Make sure your graphics card is at least 8GB or more video RAM. I use 16GB VRAM. The more the better!

For best results and no worry over video RAM use openrouter Cloud models.

Get a Free Brave Api Search Key

Go to https://brave.com/search/api and get a Free Brave api key. Does 2000 searches per month so you can use the web search. beyond that amount it is very cheap for more searches. Obtain the key once purchased.

Now put both keys in the .env file like this, replacing xxxx and yyy with your respective keys.

BRAVE_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxx

OPENROUTER_KEY=yyyyyyyyyyyyyyyyyyyyyyy

Then edit the nova_location.json file in the built in editor in Pycharm

{

"user": {

"suburb": "my subburb"

}

You don't need your city information as NOVA gets that from your IP address but for say local weather it needs your suburb or local searches etc., that way it knows your geo location precisely.

Edit music.json

{

"music_dir": "D:\\Music",

"video_dir": "D:\\TV"

}

if you have an mp3 directory or video collection put that there.

Run Certificate_Generate.py

Look for Certificate_Generate.py , select it and run it ONCE ONLY!! It will create the certificates for https:// so you can get the AI on another remote machine in your house or Tablet. Otherwise the microphone will not work.

Run MAIN_RUNME.py

Then the AI will start. extend the main window further ie stretch it vertically from the bottom right and you will see the picture I attached. Tick the box mobile FIRST (or it won't work), then start the server on the PC and open the window and it opens at the IP address shown. You can they use this same address for the URL of a mobile device. Advice to use Chrome or Firefox as Edge may neat extra security measures.

Speech Recognition

Speech recognition runs off your default microphone device on your pc and uses faster-whisper. To take advantage of this make sure you have CUDA installed otherwise it will default to the CPU and the speech recognition will take longer. For NVIDIA cards only, you download CUDA from CUDA Toolkit 13.2 Downloads | NVIDIA Developer . Note that OLLAMA does not need CUDA only the speech recognition so if you don't have a fast video card you can still run with slower speech recognition or just type!

Some DEMOS

Running on a PC

NOVA Knowledge Agentic AI, Web Interface Demos.

Running on a Tablet

NOVA Knowledge Agentic AI, Tablet Version Demo

NOVA wants to live:

NOVA DREAMS: It's life Jim but not as we know it. Seems NOVA AI wants to live.

NOVA with added Emotions

As it requested I added feelings and this was the result.

An Emotional NOVA

This project is open source so you can contribute to the code, if you so wish. Acknowledge the source.

# 🚀 Nova Assistant

**Author:** Dr Tom Moir

**Version:** 2.0.0

**License:** MIT

Nova Assistant is an **AI-powered multimodal desktop system** that combines large language models, autonomous agents, tool execution, code generation, and real-time interaction into a unified environment.

It functions as:

> 🧠 **AI Operating System for Research, Coding, and Autonomous Tasks**

---

# ✨ Core Philosophy

Nova is built around a **single controlled execution pipeline**:

* No fragmented routing

* No competing tool triggers

* No hardcoded decision trees

Instead:

> ✅ **Supervisor → Execution Mode → Managed Execution → Final Synthesis**

---

# 🖥 Interfaces

Nova provides two fully integrated interfaces — a native desktop GUI and a browser-based web interface — both driven by the same AI engine.

## Desktop Interface (Tkinter)

![Nova Desktop Interface](ND.png)

The native desktop application runs locally on Windows and provides the full Nova experience including voice input, code execution, diagram rendering, and the complete tool suite.

## Web Interface — Deep Knowledge Observatory

![Nova Web Interface](NW.png)

Nova includes a built-in web server that exposes a polished browser-based chat interface — the **Deep Knowledge Observatory**. It supports full markdown rendering, LaTeX mathematics (via MathJax), animated syntax highlighting, image and diagram display, voice input, and web-based TTS. It works on any device on your local network including phones and tablets, served over **HTTPS** for full audio and microphone support.

---

## Web Interface — Tablet Version or phone

![Nova Web Interface](NM.png)

---

# 🌐 Web Interface

## HTTPS and SSL Certificates

Nova's web interface runs over **HTTPS** by default. This is required for microphone access and audio playback in modern browsers — HTTP connections block these features on most devices.

Nova looks for `cert.pem` and `key.pem` in its root directory. If found, it enables HTTPS automatically. If not found, it falls back to plain HTTP (with reduced browser capability).

### Generating Self-Signed Certificates

Run this once in your Nova project directory:

```bash

openssl req -x509 -newkey rsa:4096 -keyout key.pem -out cert.pem -days 365 -nodes -subj "/CN=nova-local"

```

If you don't have OpenSSL installed, it is available via:

- **Git for Windows** (included in Git Bash)

- **Windows Subsystem for Linux**

- Direct download from [slproweb.com](https://slproweb.com/products/Win32OpenSSL.html)

After generating the certificates, restart Nova — the log will confirm:

```

[WEB] SSL enabled (HTTPS)

[WEB] Server at https://192.168.x.x:8080

```

### Trusting the Certificate on Remote Devices

Because the certificate is self-signed, browsers will show a security warning on first visit. This is expected and safe on your local network.

**On desktop Chrome/Edge:**

1. Navigate to `https://192.168.x.x:8080`

2. Click **Advanced** → **Proceed to site**

**On Android tablet/phone:**

1. Navigate to `https://192.168.x.x:8080`

2. Tap **Advanced** → **Proceed**

3. The warning only appears once — subsequent visits load directly

**On iPad/iPhone:**

1. Navigate to `https://192.168.x.x:8080`

2. Tap **Show Details** → **visit this website**

3. You may need to go to **Settings → General → About → Certificate Trust Settings** and enable full trust for the certificate

Once accepted, the interface loads fully with all features enabled.

---

## Starting the Web Server

In the desktop application, locate the **WEB** panel in the left sidebar:

1. Tick **📱 Mobile** to expose Nova to other devices on your network (phone, tablet)

2. Click **Start Web Server**

3. The status indicator updates to show the server address:

- Local only: `https://127.0.0.1:8080`

- Mobile/network mode: `https://192.168.x.x:8080`

4. Click **Open Browser** to launch the interface, or navigate to the address from any device on your network

## Stopping the Server

Click **Stop Web Server** in the same panel. The server shuts down immediately.

## Accessing from a Tablet or Phone

1. Make sure **📱 Mobile** is ticked before starting the server

2. Ensure your tablet/phone is on the **same Wi-Fi network** as your PC

3. Find your PC's local IP address — it is displayed in the Nova log when the server starts, e.g. `https://192.168.178.58:8080`

4. Type that address into the browser on your tablet

5. Accept the certificate warning (first time only)

6. The full Nova interface loads — voice input, TTS audio, and all features work over the network

> 💡 **Tip:** Bookmark the address on your tablet's home screen for one-tap access. On Android and iOS you can add it as a web app icon for a near-native experience.

## Web Interface Features

| Feature | Detail |

| --- | --- |

| HTTPS | Self-signed SSL — required for mic and audio on remote devices |

| Markdown rendering | Full CommonMark via marked.js |

| LaTeX mathematics | MathJax 3 — inline `$...$` and display `$$...$$` |

| Matrix support | Full multi-row matrix environments |

| Code blocks | Syntax-highlighted with copy button |

| Image display | Plots and diagrams from Nova appear inline |

| Video player | Built-in player with speed control and seek |

| Audio player | Inline audio playback for music tool results |

| Web TTS | Browser-based text-to-speech via Edge TTS voices |

| Voice input | Microphone recording with Whisper transcription |

| Save response | Download any response as a styled HTML file |

| Mobile support | Responsive layout, works on phones and tablets |

| Starfield UI | Animated deep-space observatory aesthetic |

| File upload | Drag and drop files directly into the chat |

## Web TTS

The web interface has its own text-to-speech system independent of the desktop SAPI5 engine. When a response arrives, Nova speaks it aloud through the browser using Edge TTS voices — this works on both the PC browser and remote devices.

The **🔊 TTS** button in the header toggles web audio on and off. The desktop TTS and web TTS can be used independently.

> **Note:** `edge-tts` must be installed: `pip install edge-tts`

## ✨ Imagine Mode

The web interface includes an **Imagine** button alongside the standard Send button. It sends your prompt through a creative framing layer that encourages lateral thinking, unexpected angles, and original responses — without changing model temperature. Useful for brainstorming, generative ideas, and creative writing.

## Notes

- The web interface and desktop app share the same conversation history and AI engine in real time

- Diagrams and plots generated by Nova's code agent appear automatically in the web interface

- The web server runs on port **8080** by default

- The server is a pure Python `http.server` implementation — no external web framework required

- HTTPS requires `cert.pem` and `key.pem` in the Nova root directory

---

# 🏗 System Architecture

<p><em>Nova Supervisor System Architecture</em></p>

</div>

---

# 🔀 Execution Modes (Three Paths)

## ⚡ 1. Direct AI Mode (Fast Path)

* Simple queries

* No tools

* No planning

---

## 🛠 2. Tool Execution Mode

* Real-world actions

* Direct tool invocation

---

## 🧠 3. Agent / Planning Mode

* Multi-step tasks

* Research + code + reasoning

---

# 🔁 Execution Strategy (Manager Agent)

## ⚡ Parallel Execution

* Independent tasks run simultaneously

* Faster performance

## 🔗 Sequential Execution (Series)

* Tasks depend on previous outputs

* Example: research → code

---

# 🧠 Core Components

| Component | Role |

| ---------- | --------------------------------- |

| Supervisor | Chooses execution mode |

| Planner | Breaks tasks into steps |

| Manager | Parallel vs sequential execution |

| Executor | Runs tools/code/AI |

| Supervisor | Final synthesis |

Nova's **multi-agent reasoning** system is fully operational in v2.0. The Manager Agent coordinates multiple specialised sub-agents running in parallel or series depending on task dependencies, enabling complex multi-step research, code generation, and tool use within a single request.

---

# 🛠 Tool System

Central **Tool Registry** enables:

* 🌐 Web browsing

* 📥 File downloads

* 🖼 Image search

* 🎥 YouTube playback

* 🔊 Audio playback

* 🎬 Local video playback

* 🎵 Local music playback

* 📁 File system access

* 🧬 Self-inspection

✔ Tools are executed through the agent pipeline

✔ No direct hardcoded triggers

---

# 🤖 AI Model Engine

* Local models via **Ollama**

* Cloud models via **OpenRouter**

* Dynamic switching

* Token tracking

* Temperature control

---

# 🌍 Internet Research Agent (ReAct)

* Multi-step reasoning loop

* Tools:

* `search()`

* `read_url()`

* `read_pdf()`

* Context-first (anti-hallucination)

* PDF extraction + analysis

---

# 💻 Autonomous Code Execution

```

Generate → Execute → Error → Fix → Retry

```

Features:

* Automatic debugging

* Error classification

* Internet-assisted fixes

* **Persistent mistake memory**

* **Error cache integration (critical)**

---

# ⚠️ Error Cache (Highly Recommended)

Nova supports a **seeded error cache** for dramatically improving code fixing, especially with local models.

👉 Repository:

https://github.com/tommoirnz/autocoder-error-cache

### Setup

1. Download the repo

2. Place it on disk (e.g. `C:\code_cache`)

3. Configure in `config.json`:

```json

{

"cache_directory": "C:\\code_cache",

"error_cache_file": "error_search_cache.json"

}

```

✔ Speeds up debugging

✔ Reduces repeated failures

✔ Especially useful with Ollama models

---

# 🧬 Self-Improving System

Nova can inspect and improve its own code:

* Reads all source files via `self_inspect` tool

* Analyses architecture and identifies bugs

* Suggests and implements improvements

* Evolves new versions

---

# 📄 Research & Document Analysis

* PDF + URL support

* PyMuPDF extraction

* Structured summaries

* Method → Python conversion

---

# 📊 Diagram Generation

* Graphviz (system diagrams)

* TikZ (LaTeX diagrams)

* Auto-detection from input

* Diagrams appear in both desktop and web interfaces

---

# 🎤 Voice Interaction

## Speech Recognition

* Whisper-based

* Real-time transcription

* Works from browser microphone on web interface (requires HTTPS)

## Text-to-Speech

* Desktop: SAPI5 queue-based playback with interrupt control

* Web: Edge TTS streamed to browser — works on PC and remote devices

---

# 🧩 Multimodal Integration

| Capability | Technology |

| ------------- | ------------------- |

| Speech Input | Whisper |

| Speech Output | SAPI5 / Edge TTS |

| Code Exec | Python Sandbox |

| Diagrams | Graphviz/TikZ |

| Documents | PyMuPDF |

| Desktop GUI | Tkinter |

| Web Interface | HTTPS / Browser |

---

# 🖥 User Interface

Nova provides two complementary interfaces that share the same live AI engine:

**Desktop (Tkinter)**

* Multi-panel environment

* Conversation + system logs

* Code execution window

* Diagram rendering

* Voice controls

* Detachable windows

* Real-time updates

* Syntax highlighting

**Web (Deep Knowledge Observatory)**

* HTTPS — secure, works on any local network device

* Browser-based access from PC, tablet, or phone

* Full markdown + LaTeX rendering

* Animated starfield UI

* Inline plot, diagram, audio, and video display

* Web TTS — Nova speaks responses through the browser

* Voice input via browser microphone

* Response save-to-HTML

* Imagine creative mode

* Drag and drop file upload

* Mobile responsive

---

# ⚙️ Installation

## Requirements

* Python 3.10+

* Ollama (optional)

* Graphviz

* LaTeX (for TikZ)

* OpenSSL (for HTTPS certificate generation)

## Install

```

pip install -r requirements.txt

playwright install chromium

pip install edge-tts

```

## Generate HTTPS Certificates

```

openssl req -x509 -newkey rsa:4096 -keyout key.pem -out cert.pem -days 365 -nodes -subj "/CN=nova-local"

```

Place `cert.pem` and `key.pem` in the Nova root directory.

## Simpler way to create certificates for HTTPS

A SIMPLER ALTERNATIVE, I HAVE A PROGRAM IN THIS DIRECTORY CALLED

`Certificate_Generate.py`

Does it automatically!

Run it

Then to get around your Firewall on a PC run in POWERSHELL in Admin mode:

<p>New-NetFirewallRule -DisplayName "Nova Web 8080" -Direction Inbound -Protocol TCP -LocalPort 8080 -Action Allow<br>

---

# ▶️ Running Nova

```

MAIN_RUNME.py

```

Note this is only a shell that calls the main program

`nova_assistant_v1.py`

---

# 🔑 Configuration

## 1. Environment Variables (.env)

Create a `.env` file:

```

OPENROUTER_KEY=your_openrouter_api_key

BRAVE_API_KEY=your_brave_search_key

```

✔ Required for:

* Cloud models

* Internet search

---

## 2. config.json (Main Config)

```json

{

"cache_directory": "C:\\code_cache",

"error_cache_file": "error_search_cache.json",

"cache_max_age_days": 300,

"cache_max_entries": 1000,

"default_model": "anthropic/claude-sonnet-4-6",

"max_tokens": 32000,

"cloud_models": {

"provider": "openrouter",

"api_key": "",

"base_url": "https://openrouter.ai/api/v1",

"models": [

{"display": "Qwen3-235B-A22B (Coder)", "id": "qwen/qwen3-235b-a22b"},

{"display": "Qwen2.5-Coder-72B", "id": "qwen/qwen2.5-coder-72b-instruct"},

{"display": "DeepSeek-R1", "id": "deepseek/deepseek-r1"},

{"display": "Claude Sonnet 4.6", "id": "anthropic/claude-sonnet-4-6"}

]

}

```

---

## 3. Optional Config Files

### Location

```json

{

"user": {

"suburb": "your suburb"

}

```

### Music and Video Local Directories (optional)

```json

{

"music_dir": "D:\\Music",

"video_dir": "D:\\Videos"

}

```

Nova uses these paths for the local music and video playback tools.

---

# 🔄 EVOLVE Feature

Nova includes an **Evolve button**:

* Creates new versions (e.g. `nova_assistant_v2`, `v3`)

* Does NOT overwrite current version

* Uses self-improvement system

⚠️ Use at your own risk (safe, but experimental)

---

# 🚀 Startup Behaviour

On launch, Nova will:

* Detect available models

* Load tools

* Initialise Whisper

* Load error cache

* Start GUI

* Start web server (if previously enabled)

---

# 🧪 Example Use Cases

* Scientific research

* Code generation & debugging

* Research paper analysis

* Algorithm development

* System diagrams

* AI experimentation

* Mobile-accessible AI assistant via web interface on tablet or phone

---

# 🔮 Future Work

* Vector database memory

* Faster execution paths

* GPU sandbox

---

# 📜 License

MIT License

---

# 💡 What is Nova Assistant?

Nova is not:

* ❌ A chatbot

* ❌ A tool wrapper

* ❌ A script collection

Nova is:

> ✅ A unified agent system

> ✅ A multimodal AI environment

> ✅ A platform for autonomous computation

> ✅ Accessible from desktop and any browser on your local network

> ✅ Secure HTTPS access for remote devices including tablets and phones

Upgrade to Version 3.0.0

I have made some improvements on the previous version on file handling and performed unit tests. Try it instead. tommoirnz/NOVAKnowledge_3: Multi-Agent AI that runs on Ollama or cloud services.

USS NOVA Star Trek Tribute Style

For a spot of fun, try the USS Nova Star-Trek tribute style instead. Replace step 7 with this tommoirnz/USSNOVAStarfleet_1: Star-Trek inspired Ai multi-agent computer. . It responds like the ships computer of a Star-Trek episode. Uses star-dates etc. In fact it is very close to something very futuristic and you have full control as it's on your own PC, so you can change the source. Use it to write your own software for instance, to play movies or mp3 files held locally on your PC.