Neurospicy explanation
This is the official guide to LLM, Simon Willison’s tool for sending prompts from a terminal: a window where you type computer commands. It can work with different supported models and reuse saved prompts. That becomes useful when you repeat the same task on different text. If you only want an occasional conversation, you do not need to learn a terminal first.
// Put it to use
A real-world example
You want to apply the same short editing brief to several pieces of text without retyping the brief every time.
- Read the setup section and choose a model you have access to; hosted API use may be billed separately.
- Try one short, non-sensitive paragraph with a prompt asking for a clearer version that preserves the facts.
- Follow the guide to save that editing instruction as a reusable template.
- Use the template on another paragraph and compare each answer with its original.
What you getA repeatable editing instruction you can apply from the terminal. You still review the wording and facts before using the output.
// The useful bits. The trade-offs.
Is it a good fit?
People comfortable with a terminal who want a deeper technical guide.
Pros
- See concrete examples of sending prompts and choosing models from a terminal.
- Learn how to reuse an instruction with a saved prompt template.
Cons & considerations
- Examples may make billable model calls or require extra setup.
- Reading the guide does not run a model or install its plugins.
- You will need to be comfortable following terminal instructions.
A little more detail
Learn how an explicit model choice changes a command-line workflow.
This guide covers prompt input, model selection and other ways of using LLM, Simon Willison’s command-line tool and Python library.
Read the model-selection section first, then the logging documentation before working with sensitive input. Some examples use a hosted provider. Check the chosen provider, costs and logging behaviour before running them.
// Take the idea further
What else could you do?
A reusable editing brief
Save a prompt template and apply it to a second non-sensitive paragraph.
Compare model responses
Use an available model's identifier to repeat one prompt, then judge the outputs against the same criteria.
References & credits3 sources +
Simon Willison and LLM contributors
Sources checked . Features and terms can change.