Show HN: My LLM CLI tool can run tools now, from Python code or plugins (simonwillison.net)

Article summary

The LLM CLI tool has been updated to version 0.26, which includes support for running tools in the terminal. This allows users to grant Large Language Models access to any tool that can be represented as a Python function. The tool also includes plugins, such as simpleeval and Datasette, which provide additional functionality. The update enables more complex and dynamic interactions with LLMs.

Main themes

• LLM tool updates
• tool usage and functionality
• security and safety concerns
• plugin support and customization
• research and experimentation
• comparison to other tools and libraries

What commenters say

• The LLM tool is useful for research and experimentation with different models, but its utility for everyday tasks is debated.
• Some users find the tool's ability to pipe files and outputs into an LLM to be particularly useful.
• The update increases the risk of 'footguns' and self-inflicted harm, especially when using tools that can take actions on behalf of the user.
• The tool's support for plugins and custom tools is seen as a major advantage, but also raises concerns about security and safety.
• The LLM tool is compared to other similar tools and libraries, with some users finding it more capable or convenient.
• The need for better documentation and tutorials is highlighted, particularly for users who are new to the tool or LLMs in general.