aiscout
Audit Windows AI hardware, detect local LLM runtimes like Ollama, and map model cache disk usage in under one second.
Free
Works with the AI Agents you already use
aiscout
Example session with this skill installed
Check my local system to see if I have enough VRAM to run a Llama 3 8B model and see if Ollama is currently running.
- Read your context and instructions
- Compiled the aiscout
I have scanned your local environment using aiscout. The system reports an active NVIDIA GPU with sufficient VRAM for 8B models, and Ollama was successfully detected on its default port.
AI Environment Audit
| Category | Detail | Value |
|---|---|---|
| GPU | NVIDIA GeForce RTX 3070 | 8.0 GB VRAM |
| Driver | Version 551.23 | Active |
| Runtime | Ollama (LLM Server) | Running on Port 11434 |
| Storage | Ollama Model Cache | 12.4 GB (4 models) |
| Storage | HF Hub Cache | 5.2 GB (2 models) |
| IDE | Cursor | Installed |
Model Compatibility Recommendation
Since you have 8 GB of VRAM, I recommend using llama3:8b-instruct-q4_0. Running higher precision models (Q8 or FP16) may exceed your memory capacity and cause fallback to CPU, significantly slowing down inference.
Next steps
- Pull the recommended model using
ollama pull llama3:8b-instruct-q4_0. - Configure your agent to use the local endpoint at
http://localhost:11434/v1. - Check the Hugging Face cache at
~/.cache/huggingfaceif you need to free up 5GB of space.
Connects securely to your tools. The creator never sees your data.
About this skill
Developers building local-first AI applications or managing local LLM deployments face a visibility gap when trying to determine if a machine can handle specific models. This skill audits the Windows environment to detect hardware capabilities, active runtimes, and existing model caches in under a second.
What it does
- Detect hardware specs identifying GPU models, driver versions, and total VRAM to determine model compatibility.
- Audit local runtimes finding active instances of Ollama, LM Studio, ComfyUI, and their respective ports.
- Map model caches scanning Hugging Face, Ollama, and LM Studio directories to report disk usage and model counts.
- Discover agent skills locating installed rules and configurations for Cursor, Windsurf, and Claude Code.
- Environment inspection listing available AI-related API keys and system paths without exposing sensitive values.
How it works
- Initial Scan triggered via the terminal command
npx aiscout --jsonto gather full environment data. - Hardware Analysis evaluating detected VRAM against model parameter requirements.
- Endpoint Mapping extracting base URLs and ports for local inference servers.
- Capacity Planning reporting disk usage across multiple model cache locations to manage storage.
Frameworks & tools
Specifically designed for Windows 10/11 environments running Node.js 18.0.0 or higher. It integrates with local LLM ecosystems including Ollama, LM Studio, Hugging Face, and ComfyUI.
Why this beats prompting it yourself
Prompting an LLM to guess local environment specs leads to generic or incorrect setup advice. This skill provides ground-truth JSON data, allowing the agent to make technical decisions based on actual VRAM limits and active port availability.
Use cases
- Automated model selection based on real-time VRAM capacity (e.g., picking 3B vs 8B models).
- Troubleshooting connection issues by verifying if local LLM servers like Ollama are running on expected ports.
- Managing disk space by identifying bloated Hugging Face or Ollama model caches for cleanup.
- Scaffolding agent configs by detecting existing IDE rules and local assistant installations.
Known limitations
This tool is strictly for Windows environments. It requires Node.js 18+ to be installed on the host machine to execute the scanner.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Windows only. Requires Node.js 18+. Detection is limited to predefined runtimes like Ollama, LM Studio, and ComfyUI.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 11 days ago
- Free with an Agensi account
- Software
Node.js