Ask a shell question and get a command to execute.
Example:
ollama-shell -q "put the string 'test' into a file named test.txt"
Will respond with something like this:
This command will create a new file named test.txt and write the string 'test'
into it. If the file already exists, its contents will be overwritten.
This following command has been copied to your clipboard:
echo 'test' > test.txt
pipx install git+https://github.com/diversen/ollama-shell@v0.1.1
On first usage you will need to choose a valid ollama model that exists on your system:
ollama-shell
E.g.:
Select a model from the following list:
codegeex4:latest
llama3.2:latest
codegeex4:latest will work great for shell commands.
Paste your chosen model and press return. The config file has been written and saved. On my system it is saved in the following location:
~/.local/share/ollama_shell/config.yaml
Now ask a question:
ollama-shell -q "find all json files in current folder"
The above command may return something like this:
This command searches for JSON files in the current directory and its
subdirectories. The '.' specifies the current directory, '-type f' indicates
that we are looking for files, and '-name '*.json'' filters for files with a
'.json' extension.
This following command has been copied to your clipboard:
find . -type f -name '*.json'
If you are satisfied with the answer you can press Ctrl + Shift + V to paste the command into your terminal.
ollama-shell -c
MIT