mirror of
https://github.com/RooVetGit/Roo-Code.git
synced 2026-10-02 02:11:37 +00:00
39 lines
3.5 KiB
Text
39 lines
3.5 KiB
Text
---
|
||
description: Adds a debug toggle, improves provider error messaging, and fixes tool calling issues.
|
||
keywords:
|
||
- roo code 3.39.2
|
||
- new features
|
||
- bug fixes
|
||
---
|
||
|
||
# Roo Code 3.39.2 Release Notes (2026-01-09)
|
||
|
||
This release adds a new debug toggle in settings, improves provider error messaging, and fixes multiple tool-calling issues.
|
||
|
||
## Bug Fixes
|
||
|
||
* Fixes an issue where Cerebras could fail tool calling with a 422 error when a chat included both built-in tools and MCP tools, by ensuring tools sent to Cerebras use consistent strict-mode values (thanks brianboysen51!) ([#10589](https://github.com/RooCodeInc/Roo-Code/pull/10589))
|
||
* Fixes an issue where OpenAI-compatible providers could break tool calling, causing models to describe tool usage instead of invoking tools ([#10575](https://github.com/RooCodeInc/Roo-Code/pull/10575))
|
||
* Fixes cases where file edits could fail to apply due to overly strict `edit_file` matching, improving edit reliability ([#10585](https://github.com/RooCodeInc/Roo-Code/pull/10585))
|
||
* Fixes an issue where the VS Code Language Model (vscode-lm) integration could error when tool-call parts were ordered before text parts in assistant messages ([#10573](https://github.com/RooCodeInc/Roo-Code/pull/10573))
|
||
* Fixes an issue where Gemini requests via OpenRouter could fail when an assistant message contained only tool calls ([#10559](https://github.com/RooCodeInc/Roo-Code/pull/10559))
|
||
* Fixes an issue where approving a tool run while adding feedback could produce duplicate tool results, improving reliability for unattended runs (thanks nabilfreeman!) ([#10519](https://github.com/RooCodeInc/Roo-Code/pull/10519))
|
||
* Fixes an issue where Gemini tool calling could behave inconsistently across turns due to thought signatures not being preserved correctly ([#10590](https://github.com/RooCodeInc/Roo-Code/pull/10590))
|
||
|
||
## Provider Updates
|
||
|
||
* Improves file-edit reliability for Gemini and Vertex models by disabling the `edit_file` tool for those providers so Roo can route edits through the more appropriate patch-based flow when available ([#10594](https://github.com/RooCodeInc/Roo-Code/pull/10594))
|
||
* Improves Gemini tool selection behavior by no longer overriding tool allow/deny lists, letting Roo’s routing decide more consistently ([#10592](https://github.com/RooCodeInc/Roo-Code/pull/10592))
|
||
|
||
## QOL Improvements
|
||
|
||
* Adds an **Enable debug mode** toggle in Roo Code settings (About Roo Code) so you can turn debug logging on/off from the extension UI ([#10580](https://github.com/RooCodeInc/Roo-Code/pull/10580))
|
||
* Improves the error message shown when a provider terminates a streaming response early (for example timeouts or rate limits), making it clearer the interruption came from the provider ([#10548](https://github.com/RooCodeInc/Roo-Code/pull/10548))
|
||
|
||
## Misc Improvements
|
||
|
||
* Improves CLI defaults by switching the default model to `anthropic/claude-opus-4.5` ([#10544](https://github.com/RooCodeInc/Roo-Code/pull/10544))
|
||
* Updates Roo Code Cloud’s Terms of Service (effective 2026-01-09) ([#10568](https://github.com/RooCodeInc/Roo-Code/pull/10568))
|
||
* Improves internal CLI development and maintenance by moving more shared types to `@roo-code/types` (thanks cte!) ([#10583](https://github.com/RooCodeInc/Roo-Code/pull/10583))
|
||
* Improves internal CLI stability by adding new functionality to `@roo-code/core` (thanks cte!) ([#10584](https://github.com/RooCodeInc/Roo-Code/pull/10584))
|
||
* Improves internal CLI workflows by adding additional slash commands for CLI development (thanks cte!) ([#10586](https://github.com/RooCodeInc/Roo-Code/pull/10586))
|