Step‑by‑Step Guide to Install Ollama and ShellGPT for Local LLM Use

Install Ollama (Windows)

Step 1: Download Ollama

Visit https://ollama.com/download to download the Windows installer.

Step 2: Search on Ollama website

Open the Ollama homepage to locate the desired model.

Step 3: Deploy in terminal

Install the model that matches your computer configuration.

ollama run gemma2:2b

Install ShellGPT on Linux

Step 1: Install ShellGPT

ShellGPT uses LiteLLM to communicate with the local Ollama backend. Install it with:

pip install shell-gpt --break-system-packages

Step 2: Edit .sgptrc configuration

Edit ~/.config/shell_gpt/.sgptrc and add options such as:

DEFAULT_MODEL=ollama/gemma2:9b
OPENAI_USE_FUNCTIONS=false
SHOW_FUNCTIONS_OUTPUT=false
API_BASE_URL=http://192.168.79.249:11434
# other settings omitted for brevity

Initialize ShellGPT

Provide any string for the initial API key, then run:

sgpt

ShellGPT functional tests

Example of asking “Who are you?”: sgpt "你是谁" ShellGPT replies as an AI assistant for programming and system management.

Example of generating a shell command: sgpt -s "查看当前目录" ShellGPT suggests using pwd and explains the reasoning.

Generate code example:

sgpt -c "写一个shell脚本，要求获取cpu核数线程，内存大小，磁盘大小，网络带宽"

List chats, show chat info, start REPL:

sgpt --list-chats
sgpt --show-chat <chat_path>
sgpt --repl temp