VincentAtHome

Local Windows PowerShell + Ollama file-organiser harness (plain English in, validated moves out).

Windows PowerShell scripts that pair plain-English prompts with a local Ollama model (for example gemma4:e2b) to plan folder tidy-up operations. The harness resolves paths, shows a snapshot, asks questions first, validates every action, supports dry-run, and only runs moves after you type y.

Prerequisites

• Windows with PowerShell
• Ollama installed and running on 127.0.0.1:11434
• At least one Google Gemma model pulled locally (this project defaults to gemma4:e2b)

Install Ollama and Google Gemma models

Install Ollama (Windows)

1. Open Ollama — Download and download the Windows installer.

2. Run the installer and complete the steps (you may need to allow the app through the firewall the first time it serves models).

3. After installation, Ollama usually listens on http://127.0.0.1:11434. You do not need to run ollama serve manually if the Ollama app or background service is already running (starting it twice can show a “bind: Only one usage…” message, which means the port is already in use).

4. Check that the CLI works:

   ollama --version

5. Optional sanity check that the HTTP API responds:

   Invoke-RestMethod -Uri "http://127.0.0.1:11434/api/tags" -Method Get

Install Google Gemma models

Gemma models in Ollama are listed in the Ollama library (search for “gemma”). Tags and sizes change over time; always pick a tag that exists on your machine with ollama list.

Pull models (downloads weights; needs disk space and a stable connection):

# Default for this harness (see $MODEL in ai/ai.ps1)
ollama pull gemma4:e2b

# Smaller / faster option for testing (optional)
ollama pull gemma3:1b

Confirm they are installed:

ollama list

Using a different model: edit the $MODEL line near the top of ai/ai.ps1 to match a name shown in ollama list (for example gemma3:1b). If you use ai-status.ps1 warm-up, update the ollama run … line there so it matches the same tag.

Licence and terms: Gemma is provided by Google under its own terms. Review Google’s Gemma documentation and terms before use, especially for redistribution or commercial use.

Install this harness

1. Clone this repository (or copy the ai folder).

2. Copy the ai folder to a fixed location, for example:

   C:\ai\

   so you have:

   • C:\ai\ai.ps1
   • C:\ai\ai-status.ps1

3. Optional: define commands in your PowerShell profile ($PROFILE). Aliases cannot point at .ps1 paths directly, so use functions (see install/profile-snippet.ps1):

   $AiRoot = "C:\ai"
   function ai { & (Join-Path $AiRoot "ai.ps1") @args }
   function aistatus { & (Join-Path $AiRoot "ai-status.ps1") @args }

   Adjust $AiRoot if you keep the scripts elsewhere.

Usage

Check Ollama and models:

aistatus

Organise a folder (examples):

ai "tidy my desktop" -DryRun
ai "organise my downloads" -DryRun
cd "D:\Your\Project"
ai "tidy this folder" -DryRun
ai "can you tidy this folder?" -TargetPath "D:\Your\Exact\Folder" -DryRun

Remove -DryRun only when you intend to apply changes; the script will ask for final confirmation.

Behaviour (summary)

• Type mode: deterministic moves by file extension (no model-generated filenames).
• Purpose / date mode: the model returns JSON classifications only; the script builds New-Item / Move-Item lines and checks every filename against the live snapshot.
• Safety checks include allowlisted cmdlets, quoted paths, staying inside the target folder, and no deletes.

Repository layout

Path	Purpose
assets/vincent-at-home.png	README mascot image
ai/ai.ps1	Main harness
ai/ai-status.ps1	Ollama reachability, ollama list, optional model warm-up
install/profile-snippet.ps1	Example profile functions for ai and aistatus

Local-only folder (not on GitHub)

Create a social/ folder next to ai/ if you want Markdown snapshots of X.com or other threads. It is listed in .gitignore, so it stays on your machine and is never committed or pushed to the remote repository.

Licence

Use and modify for personal or internal use; add a licence file if you redistribute.