OpenClaw v2026.3.2 introduces PDF analysis tool and 150+ fixes

OpenClaw has unveiled version v2026.3.2, which introduces a built-in PDF analysis tool and over 150 bug fixes, marking a significant update for the open-source AI agent framework. This release, dated March 3, 2026, showcases contributions from 93 developers and includes enhancements such as a new Speech-to-Text API for audio transcription and improved Telegram message streaming capabilities. Notably, the PDF analysis tool supports processing backends from Anthropic and Google, allowing for configurable extraction policies and page limits.

In addition to new features, this update also brings several breaking changes, including a shift in the default tool configuration for new installations from a broad programming toolset to a messaging-focused setup. Developers utilizing OpenClaw will benefit from improved security mechanisms, such as the extended SecretRef credential reference system, which now accommodates 64 targets. With these updates, OpenClaw continues to enhance its developer tooling and API integrations, ensuring a robust agent workflow for users.