llm-anthropic 0.27 Integrates New Anthropic Python SDK

Developer Simon Willison has officially released version 0.27 of his llm-anthropic plugin, a tool designed to grant command-line access to Anthropic's Claude series of large language models. Announced on August 24, 2026, this update primarily introduces compatibility with the newly launched anthropic v1.0.0 Python library. The core change in this major SDK release is a transition from the httpx library to httpx2, mirroring a similar architectural shift made by OpenAI in its own v3.0.0 library release two weeks earlier.

To implement the upgrade, Willison leveraged automated coding tools rather than manual refactoring. He utilized the Fable 5 model within Claude Code, prompting the AI agent to read Anthropic's official migration guide hosted on GitHub and update the codebase to support version 1.0 or higher while ensuring all existing tests passed. This workflow highlights the growing utility of agentic coding assistants in maintaining open-source software ecosystems when upstream dependencies shift.

For developers and practitioners relying on the LLM CLI tool, this release ensures uninterrupted access to Anthropic's model suite without dependency conflicts. As major AI providers transition to newer HTTP clients like httpx2, maintaining alignment across different model plugins becomes critical. The update prevents version lock-ins and allows developers to run the latest versions of both OpenAI and Anthropic libraries side-by-side in their local environments.

The transition to httpx2 across major AI SDKs represents a broader push toward modernizing the networking stack for LLM integrations. By keeping the llm-anthropic plugin updated, Willison ensures that command-line users can continue to leverage Claude's capabilities with the performance and security improvements offered by the updated HTTP client. Practitioners can install or upgrade the plugin immediately to benefit from these underlying library enhancements.