feat: add GPT-5-Codex support and keyboard shortcut for auto-approve in v3.28.6 release notes (#361)

hannesrudolph · web-flow · commit c1fd9117eb48 · 2025-09-23T12:50:05.000-06:00
diff --git a/docs/features/auto-approving-actions.mdx b/docs/features/auto-approving-actions.mdx
@@ -43,6 +43,24 @@ Auto-approve settings speed up your workflow by eliminating repetitive confirmat
 3. Use the All/None chips to bulk-select or clear permissions, or select individual tiles; you can keep Enabled On with "None" selected
 4. (Optional) Click the gear icon to open Settings for deeper per-permission controls
 
+### Keyboard Shortcut
+
+**Default shortcut:** `Cmd+Alt+A` (macOS) / `Ctrl+Alt+A` (Windows/Linux)
+
+Quickly toggle auto-approve on/off without using the mouse. This shortcut toggles the global "Enabled" state while preserving your permission selections.
+
+**To customize the shortcut:**
+1. Open VS Code Command Palette (`Cmd+Shift+P` / `Ctrl+Shift+P`)
+2. Search for "Preferences: Open Keyboard Shortcuts"
+3. Search for the command name (varies by language):
+   - English: "Toggle Auto-Approve"
+   - Other languages: Look for the localized equivalent
+4. Click the pencil icon next to the command
+5. Press your desired key combination
+6. Press Enter to save
+
+**Note:** The command name appears in your VS Code interface language. If you're using a non-English locale, the command will be translated accordingly.
+
 ---
 
 ## Auto-Approve Dropdown
diff --git a/docs/providers/ollama.md b/docs/providers/ollama.md
@@ -53,8 +53,14 @@ Roo Code supports running models locally using Ollama. This provides privacy, of
     ollama pull qwen2.5-coder:32b
     ```
 
-3. **Configure the Model:** Configure your model’s context window in Ollama and save a copy. Roo automatically reads the model’s reported context window from Ollama and passes it as `num_ctx`; no Roo-side context size setting is required for the Ollama provider.
+3. **Configure the Model:** Configure your model's context window in Ollama and save a copy.
 
+   :::info Default Context Behavior
+   **Roo Code automatically defers to the Modelfile's `num_ctx` setting by default.** When you use a model with Ollama, Roo Code reads the model's configured context window and uses it automatically. You don't need to configure context size in Roo Code settings - it respects what's defined in your Ollama model.
+   :::
+
+   **Option A: Interactive Configuration**
+   
    Load the model (we will use `qwen2.5-coder:32b` as an example):
    
     ```bash
@@ -73,6 +79,37 @@ Roo Code supports running models locally using Ollama. This provides privacy, of
     /save your_model_name
     ```
 
+   **Option B: Using a Modelfile (Recommended)**
+
+   Create a `Modelfile` with your desired configuration:
+
+   ```dockerfile
+   # Example Modelfile for reduced context
+   FROM qwen2.5-coder:32b
+   
+   # Set context window to 32K tokens (reduced from default)
+   PARAMETER num_ctx 32768
+   
+   # Optional: Adjust temperature for more consistent output
+   PARAMETER temperature 0.7
+   
+   # Optional: Set repeat penalty
+   PARAMETER repeat_penalty 1.1
+   ```
+
+   Then create your custom model:
+
+   ```bash
+   ollama create qwen-32k -f Modelfile
+   ```
+
+   :::tip Override Context Window
+   If you need to override the model's default context window:
+   - **Permanently:** Save a new model version with your desired `num_ctx` using either method above
+   - **Roo Code behavior:** Roo automatically uses whatever `num_ctx` is configured in your Ollama model
+   - **Memory considerations:** Reducing `num_ctx` helps prevent out-of-memory errors on limited hardware
+   :::
+
 4.  **Configure Roo Code:**
     *   Open the Roo Code sidebar (<KangarooIcon /> icon).
     *   Click the settings gear icon (<Codicon name="gear" />).
@@ -120,17 +157,22 @@ If no model instance is running, Ollama spins one up on demand. During that cold
      /set parameter num_ctx 32768
      /save &lt;your_model_name&gt;
      ```
-   - Option B — Modelfile:
-     ```text
+   - Option B — Modelfile (recommended for reproducibility):
+     ```dockerfile
+     FROM &lt;base-model&gt;
      PARAMETER num_ctx 32768
+     # Adjust based on your available memory:
+     # 16384 for ~8GB VRAM
+     # 32768 for ~16GB VRAM
+     # 65536 for ~24GB+ VRAM
      ```
-     Then re-create the model:
+     Then create the model:
      ```bash
      ollama create &lt;your_model_name&gt; -f Modelfile
      ```
 
 3. **Ensure the model's context window is pinned**
-   Save your Ollama model with an appropriate `num_ctx` (e.g., via `/set` + `/save`, or a Modelfile). Roo reads this automatically and passes it as `num_ctx`; there is no Roo-side context size setting for the Ollama provider.
+   Save your Ollama model with an appropriate `num_ctx` (via `/set` + `/save`, or preferably a Modelfile). **Roo Code automatically detects and uses the model's configured `num_ctx`** - there is no manual context size setting in Roo Code for the Ollama provider.
 
 4. **Use smaller variants**
    If GPU memory is limited, use a smaller quant (e.g., q4 instead of q5) or a smaller parameter size (e.g., 7B/13B instead of 32B).
diff --git a/docs/providers/openai.md b/docs/providers/openai.md
@@ -43,6 +43,17 @@ The GPT-5 models are OpenAI's most advanced, offering superior coding capabiliti
 * **`gpt-5-mini-2025-08-07`** - Faster, cost-efficient for well-defined tasks
 * **`gpt-5-nano-2025-08-07`** - Fastest, most cost-efficient option
 
+### GPT-5-Codex
+OpenAI's specialized coding model with advanced capabilities:
+
+**Key Features:**
+* **400K Token Context Window** - Process entire codebases and lengthy documentation
+* **Image Support** - Analyze screenshots, diagrams, and visual documentation
+* **Prompt Caching** - Reduced costs for repeated context through automatic caching
+* **Adaptive Reasoning** - Dynamically adjusts reasoning depth based on task complexity
+
+**Ideal for:** Large-scale code analysis, multimodal tasks requiring visual understanding, complex refactoring projects, and extensive codebase operations.
+
 ### GPT-4.1 Family
 Advanced multimodal models with balanced capabilities:
 
@@ -136,5 +147,6 @@ GPT-5 models maintain conversation context efficiently through response IDs, red
 
 ## Tips and Notes
 
-*   **Pricing:** Refer to the [OpenAI Pricing](https://openai.com/pricing) page for details on model costs.
+*   **Pricing:** Refer to the [OpenAI Pricing](https://openai.com/pricing) page for current model costs and discounts, including prompt caching.
 *   **Azure OpenAI Service:** If you'd like to use the Azure OpenAI service, please see our section on [OpenAI-compatible](/providers/openai-compatible) providers.
+*   **Context Optimization:** For GPT-5-Codex, leverage prompt caching by maintaining consistent context across requests to reduce costs significantly.
diff --git a/docs/update-notes/index.md b/docs/update-notes/index.md
@@ -20,6 +20,7 @@ image: /img/social-share.jpg
 ### Version 3.28
 
 *   [3.28](/update-notes/v3.28) (Combined)
+*   [3.28.6](/update-notes/v3.28.6) (2025-09-23)
 *   [3.28.5](/update-notes/v3.28.5) (2025-09-20)
 *   [3.28.4](/update-notes/v3.28.4) (2025-09-19)
 *   [3.28.3](/update-notes/v3.28.3) (2025-09-16)
diff --git a/docs/update-notes/v3.28.6.mdx b/docs/update-notes/v3.28.6.mdx
@@ -0,0 +1,38 @@
+---
+description: GPT-5-Codex arrives in OpenAI Native alongside localization tooling and UI refinements.
+keywords:
+  - roo code 3.28.6
+  - gpt-5-codex
+  - localization
+  - bug fixes
+  - release notes
+image: /img/social-share.jpg
+---
+
+# Roo Code 3.28.6 Release Notes (2025-09-23)
+
+This release adds GPT-5-Codex to OpenAI Native, sharpens localization coverage, and smooths UI workflows across languages.
+
+## GPT-5-Codex lands in OpenAI Native
+
+- **Work with repository-scale context**: Keep multi-file specs and long reviews in a single thread thanks to a 400k token window.
+- **Reuse prompts faster and include visuals**: Prompt caching and image support help you iterate on UI fixes without re-uploading context.
+- **Let the model adapt its effort**: GPT-5-Codex automatically balances quick responses for simple questions with deeper reasoning on complex builds.
+
+This gives teams a higher-capacity OpenAI option without extra configuration.[#8260](https://github.com/RooCodeInc/Roo-Code/pull/8260):
+
+
+> **📚 Documentation**: See [OpenAI Provider Guide](/providers/openai) for capabilities and setup guidance.
+
+## QOL Improvements
+
+* **Keyboard shortcut for auto-approve**: Toggle approvals with Cmd/Ctrl+Alt+A from anywhere in the editor, keeping focus on the code review flow (via [#8214](https://github.com/RooCodeInc/Roo-Code/pull/8214))
+* **Cleaner code blocks**: Removed the snippet language picker and word-wrap toggle so wrapped code is easier to read and copy across locales (via [#8208](https://github.com/RooCodeInc/Roo-Code/pull/8208))
+* **More readable reasoning blocks**: Added spacing before section headers inside reasoning transcripts to make long explanations easier to scan (via [#7868](https://github.com/RooCodeInc/Roo-Code/pull/7868))
+* **Translation checks cover package settings**: The missing translation finder now validates package.nls files for 17 locales to catch untranslated VS Code strings earlier (via [#8255](https://github.com/RooCodeInc/Roo-Code/pull/8255))
+
+## Bug Fixes
+
+* **Bare-metal evals stay signed in**: Roo provider tokens refresh automatically and the local evals app binds to port 3446 for predictable scripts (via [#8224](https://github.com/RooCodeInc/Roo-Code/pull/8224))
+* **Checkpoint text stays on one line**: Prevented multi-line wrapping in languages such as Chinese, Korean, Japanese, and Russian so the checkpoint UI stays compact (via [#8207](https://github.com/RooCodeInc/Roo-Code/pull/8207); reported in [#8206](https://github.com/RooCodeInc/Roo-Code/issues/8206))
+* **Ollama respects Modelfile num_ctx**: Roo now defers to your Modelfile’s context window to avoid GPU OOMs while still allowing explicit overrides when needed (via [#7798](https://github.com/RooCodeInc/Roo-Code/pull/7798); reported in [#7797](https://github.com/RooCodeInc/Roo-Code/issues/7797))
diff --git a/docs/update-notes/v3.28.mdx b/docs/update-notes/v3.28.mdx
@@ -52,14 +52,27 @@ Task Sync enables monitoring your local development environment from any device.
 
 > **Documentation**: See [Task Sync](/roo-code-cloud/task-sync), [Roomote Control Guide](/roo-code-cloud/roomote-control), and [Billing & Subscriptions](/roo-code-cloud/billing-subscriptions).
 
+## GPT-5-Codex lands in OpenAI Native
+
+- **Work with repository-scale context**: Keep multi-file specs and long reviews in a single thread thanks to a 400k token window.
+- **Reuse prompts faster and include visuals**: Prompt caching and image support help you iterate on UI fixes without re-uploading context.
+- **Let the model adapt its effort**: GPT-5-Codex automatically balances quick responses for simple questions with deeper reasoning on complex builds.
+
+This gives teams a higher-capacity OpenAI option without extra configuration.[#8260](https://github.com/RooCodeInc/Roo-Code/pull/8260):
+
+> **Documentation**: See [OpenAI Provider Guide](/providers/openai) for capabilities and setup guidance.
+
 ## QOL Improvements
 
+* **Auto-approve keyboard shortcut**: Toggle approvals with Cmd/Ctrl+Alt+A from anywhere in the editor so you can stay in the flow while reviewing changes (via [#8214](https://github.com/RooCodeInc/Roo-Code/pull/8214))
 * **Click-to-Edit Chat Messages**: Click directly on any message text to edit it, with ESC to cancel and improved padding consistency ([#7790](https://github.com/RooCodeInc/Roo-Code/pull/7790))
 * **Enhanced Reasoning Display**: The AI's thinking process now shows a persistent timer and displays reasoning content in clean italic text ([#7752](https://github.com/RooCodeInc/Roo-Code/pull/7752))
+* **Easier-to-scan reasoning transcripts**: Added clear line breaks before reasoning headers inside the UI so long thoughts are easier to skim (via [#7868](https://github.com/RooCodeInc/Roo-Code/pull/7868))
 * **Manual Auth URL Input**: Users in containerized environments can now paste authentication redirect URLs manually when automatic redirection fails ([#7805](https://github.com/RooCodeInc/Roo-Code/pull/7805))
 * **Active Mode Centering**: The mode selector dropdown now automatically centers the active mode when opened ([#7883](https://github.com/RooCodeInc/Roo-Code/pull/7883))
 * **Preserve First Message**: The first message containing slash commands or initial context is now preserved during conversation condensing instead of being replaced with a summary ([#7910](https://github.com/RooCodeInc/Roo-Code/pull/7910))
 * **Checkpoint Initialization Notifications**: You'll now receive clear notifications when checkpoint initialization fails, particularly with nested Git repositories ([#7766](https://github.com/RooCodeInc/Roo-Code/pull/7766))
+* **Translation coverage auditing**: The translation checker now validates package.nls locales by default to catch missing strings before release (via [#8255](https://github.com/RooCodeInc/Roo-Code/pull/8255))
 
 * Smaller and more subtle auto-approve UI (thanks brunobergher!) ([#7894](https://github.com/RooCodeInc/Roo-Code/pull/7894))
 * Disable Roomote Control on logout for better security ([#7976](https://github.com/RooCodeInc/Roo-Code/pull/7976))
@@ -74,10 +87,13 @@ Task Sync enables monitoring your local development environment from any device.
 * **Redesigned Message Feed**: Enjoy a cleaner, more readable chat interface with improved visual hierarchy that helps you focus on what matters ([#7985](https://github.com/RooCodeInc/Roo-Code/pull/7985))
 * **Responsive Auto-Approve**: The auto-approve dropdown now adapts to different window sizes with smart 1-2 column layouts, and tooltips show all enabled actions without truncation ([#8032](https://github.com/RooCodeInc/Roo-Code/pull/8032))
 * **Network Resilience**: Telemetry data now automatically retries on network failures, ensuring analytics and diagnostics aren't lost during connectivity issues ([#7597](https://github.com/RooCodeInc/Roo-Code/pull/7597))
-* **Code blocks wrap by default**: Code blocks now wrap text by default, improving readability when viewing long commands and code snippets ([#8194](https://github.com/RooCodeInc/Roo-Code/pull/8194))
+* **Code blocks wrap by default**: Code blocks now wrap text by default, and the snippet toolbar no longer includes language or wrap toggles, keeping snippets readable across locales (via [#8194](https://github.com/RooCodeInc/Roo-Code/pull/8194); [#8208](https://github.com/RooCodeInc/Roo-Code/pull/8208))
 
 ## Bug Fixes
 
+* **Roo provider stays signed in**: Roo provider tokens refresh automatically and the local evals app binds to port 3446 for predictable scripts (via [#8224](https://github.com/RooCodeInc/Roo-Code/pull/8224))
+* **Checkpoint text stays on one line**: Prevented multi-line wrapping in languages such as Chinese, Korean, Japanese, and Russian so the checkpoint UI stays compact (via [#8207](https://github.com/RooCodeInc/Roo-Code/pull/8207); reported in [#8206](https://github.com/RooCodeInc/Roo-Code/issues/8206))
+* **Ollama respects Modelfile num_ctx**: Roo now defers to your Modelfile’s context window to avoid GPU OOMs while still allowing explicit overrides when needed (via [#7798](https://github.com/RooCodeInc/Roo-Code/pull/7798); reported in [#7797](https://github.com/RooCodeInc/Roo-Code/issues/7797))
 * **Groq Context Window**: Fixed incorrect display of cached tokens in context window ([#7839](https://github.com/RooCodeInc/Roo-Code/pull/7839))
 * **Chat Message Operations**: Resolved duplication issues when editing messages and "Couldn't find timestamp" errors when deleting ([#7793](https://github.com/RooCodeInc/Roo-Code/pull/7793))
 * **UI Overlap**: Fixed CodeBlock button z-index to prevent overlap with popovers and configuration panels (thanks A0nameless0man!) ([#7783](https://github.com/RooCodeInc/Roo-Code/pull/7783))
diff --git a/sidebars.ts b/sidebars.ts
@@ -223,6 +223,7 @@ const sidebars: SidebarsConfig = {
           label: '3.28',
           items: [
             { type: 'doc', id: 'update-notes/v3.28', label: '3.28 Combined' },
+            { type: 'doc', id: 'update-notes/v3.28.6', label: '3.28.6' },
             { type: 'doc', id: 'update-notes/v3.28.5', label: '3.28.5' },
             { type: 'doc', id: 'update-notes/v3.28.4', label: '3.28.4' },
             { type: 'doc', id: 'update-notes/v3.28.3', label: '3.28.3' },
