llm-coding-agent 0.1a0

Simon Willison released llm-coding-agent 0.1a0, an experimental coding agent built on his evolving LLM library. The agent follows a Claude Code-inspired design, providing tools for file reading, editing, and command execution. Willison developed it by first writing a spec, then building it using red/green TDD in a series of commits, each with passing tests and updated docs. The agent is available on PyPI and can be run via `uvx --prerelease=allow --with llm-coding-agent llm code`. It supports both a CLI with flags like `--yolo` and a Python API using the `CodingAgent` class. The Python API, which Willison did not explicitly request but appreciates, allows developers to programmatically define an agent with a model, root directory, and approval mode. The agent also exposes a suite of tools accessible via `uvx ... llm tools`. This release demonstrates a practical approach to building a simple coding agent using an agent framework, leveraging existing LLM capabilities and TDD methodology.