A shell utility that allows you to generate shell command syntax for any scenario, specified by you, in natural language (currently english). Powered by a local LLM API (Ollama) and a language model (qwen2.5-coder).
llsh <natural language query for a shell command>
Example:
llsh list all files in current directory
- Generate response for command line syntax query
- Get environment variables from OS.
- Clean natural language query.
- Pass natural language query to Ollama + Language Model.
- Get response from Ollama + Language Model.
- Clean response.
- Return response to user to execute.
- Allow user to specify API server from environment variable.
- Allow user to specify model from environment variable or some other mechanism.
- Get environment variables from OS.
- Allow auto-completion of command
- Allow parsing through arguments of a command
- Prefix a long complicated input command with our program and press return key.
- Our program parses the command for arguments.
- Step-wise allow the user to modify each argument.
- Auto generate other possible outcomes for each argument, using AI.
- Add RAG & Internet Search.
- More to come...
- Ollama + Suitable Model: For running language model locally.
- Open-WebUI [ToDO]: Cleaner interface to Ollama & the model.