Simon Willison Releases llm 0.31.1 with Tool Call Bug Fix

Developer Simon Willison released version 0.31.1 of his open-source tool llm — a command-line utility and Python library for working with large language models through the terminal.

What exactly was fixed

The patch eliminates a bug in the OpenAI Chat Completion endpoints: when a model invoked a tool call with empty arguments, some providers responded with a JSON format error instead of a correct result. The error was recorded in issue #1521 of the project repository on GitHub.

How the bug was found

Willison discovered the problem while testing another of his projects — the llm-meta-ai plugin, which adds support for Meta models to llm. The llm tool is designed as a plugin system: the basic CLI can be extended with support for new providers and models without changing the core, so bugs in the basic logic of tool invocation affect all connected plugins at once.

The project's author is also known as the creator of Datasette — an open tool for publishing and exploring data, so his work in LLM tooling has long been of interest to the developer community.

What this means

Even small CLI tools for working with LLMs increasingly rely on the reliability of the tool calling mechanism — invoking external functions by the model. JSON format errors in such chains can break entire automated pipelines, so such targeted patches are important for developers building agents on top of third-party APIs.