Skip to main content

Roo Code 3.29.1 Release Notes (2025-10-27)

This patch prevents context overruns on long generations, restores the standard Redo keybinding, and reduces intermittent provider model loading errors.

Roo Code v3.29.1 Release

QOL Improvements​

  • Keyboard shortcut update: “Add to Context” moved to Ctrl+K Ctrl+A (Windows/Linux) / Cmd+K Cmd+A (macOS) to avoid Redo conflict and restore standard Redo (thanks swythan!) (#8653)

Bug Fixes​

  • Clean up max output token calculations to prevent context window overruns (thanks enerage!) (#8822)
  • Fix provider model loading race conditions to reduce timeouts and intermittent errors (#8836)

Provider Updates​

  • OpenAI-compatible: centralized ~20% maxTokens cap to prevent context overruns; GLM‑4.6‑turbo default 40,960 for reliable long-context runs (thanks enerage!) (#8822)