0.34.0

Highlights

✨ Fine-Grained Model Control with Provider Overrides

Take unprecedented control over your AI models! We've introduced provider overrides at the model level, allowing you to set specific parameters that your chosen provider offers. This means you can now, for example, configure an OpenRouter model to prioritize specific underlying providers or order them by price. For Gemini models, you could set a specific thinking budget for individual models, giving you precise control over their behavior and cost efficiency.

⚡ Performance Boosts

We've made several key improvements to enhance AiderDesk's performance:

• Faster Model Library: The Model Library now uses a virtualized table, dramatically improving performance when managing a large number of available models.
• Experimental Message Virtualization: For users experiencing lag with many messages and diff viewers, we've added experimental support for virtualized rendering of messages. This feature, disabled by default, can be enabled in Settings -> General -> Message Display. While there are some known issues we're actively working on, we encourage you to try it out and
  provide feedback! You'll also notice the Markdown rendering option has moved to this global setting for easier access.

• Optimized File Operations: The glob and grep tools have received a performance upgrade by more efficiently handling ignored files using git check-ignore.

📊 Markdown Tables Rendering

Enhance the readability of your AI's outputs! AiderDesk now supports rendering Markdown tables within messages, making structured data and summaries much clearer and easier to digest.

🚀 Direct Support for Z.AI Coding Plan Provider

We've added direct support for the Z.AI Coding Plan provider. If you're looking for a powerful and cost-efficient model for your coding plans, you can now configure and utilize Z.AI directly within AiderDesk without any workarounds.

Other Changes

• Environment variables from Settings > Aider are now correctly used when loading provider models.
• The AIDER_DESK_DATA_DIR environment variable is now properly respected for custom data directories.
• Added support for an ARGUMENTS placeholder in custom commands for more dynamic workflows.
• Reloading a project now preserves and keeps your current session active.
• Updated to the latest Electron version to ensure compatibility and resolve issues with the newest macOS releases.
• Improved the subagent run task tool to use name-based IDs instead of UUIDs, preventing the wrong subagent from being executed.
• Thinking messages are no longer added after interrupting a tool call, providing a cleaner experience.

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.33.0

Highlights

✨ Introducing the Unified Model Library: Smarter Model Management

We're excited to introduce the Model Library, a significant step forward in simplifying and unifying how you manage AI models across AiderDesk. This major refactor streamlines model selection, eliminating the need to remember provider prefixes (like requesty/ or openai/) and providing a consistent experience whether you're in Agent, Code, Ask, or Architect mode.

Key enhancements include:

• Unified Selection: A single, intuitive interface for choosing models, regardless of the active mode.
• Multiple Provider Profiles: Configure multiple profiles for the same LLM provider (e.g., separate Anthropic accounts for personal and work, or various OpenAI-compatible providers with custom prefixes like zai/).
• Automatic & Custom Models: Models are now automatically loaded from your configured providers. You can also easily add your own custom models if they aren't detected or need specific configurations.
• Enhanced Metadata: We've updated the way model metadata, such as token limits and cost per token, is loaded, ensuring more accurate information. You can also manually set these values for any model.

This extensive change aims to make your interaction with AiderDesk's AI models more seamless and powerful. While we've thoroughly tested many workflows, given the scope of these changes, we encourage you to report any specific issues you might encounter. Stay tuned, as there's much more to come for the Model Library!

☁️ Azure Provider Support

AiderDesk now officially supports Azure OpenAI Service! You can seamlessly integrate your Azure-deployed models, expanding your options for secure and scalable AI interactions within your enterprise environment.

Other Changes

• Fixed an issue where the diff edit format was not being set correctly on startup.
• You can now customize the AIDER_DESK_DATA_DIR via an environment variable for more flexible data management.
• Resolved an issue where it wasn't possible to interrupt an agent run when switching between projects.
• Fixed a bug with context files not being highlighted in "All files" mode on Windows.
• Improved the parsing of thinking blocks to prevent nested display issues.
• Added more actions for project tabs, enhancing project management. (Thanks @kristian-ziga)

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.32.1

Hotfix

• reverted yes-allways option behavior

This change has been reverted because it broke the confirmation flow in Aider. Given how this option is handled within Aider's source code, a different approach will be required.

Complete release notes for version 0.32.0 are available here.

0.32.0

Highlights

✨ Enhanced Subagent Message Visibility

Stay even more informed about your subagents' progress! We've improved the visibility of subagent messages, allowing you to see compact information about their current actions directly within the main chat, even without expanding the subagent group message. This provides a clearer, real-time understanding of what each specialized assistant is doing.

🧪 Experimental XML System Prompt Structure for Better Instruction Following

We're introducing an experimental change to the system prompt structure, now utilizing XML format. This enhancement is designed to improve how various AI models interpret and follow instructions, leading to more consistent and accurate responses. While we expect this to improve performance across most models, please be aware that in some cases, you might experience behavior different from what you're used to. If you notice any unexpected changes, please let us know – your feedback is crucial!

⚡ New Provider: Cerebras

AiderDesk now officially supports Cerebras as a new LLM provider! (Thank you to @Arefinw for the suggestion!)

🧹 clear_context Tool Added to AiderDesk MCP Server

For those leveraging the AiderDesk MCP server with other clients, we've added a new clear_context tool. This allows for more explicit control over the AI's context. If you have suggestions for more tools you'd like to see added to the MCP server, feel free to ask as @giantbirdicus did here!

Other Changes

• Fixed an annoying issue where text would jump in the Prompt field when typing quickly.
• Notifications now correctly adapt to your selected theme for a more consistent UI. (Thanks @kristian-ziga)
• Added checks for project paths to ensure only directories are handled. (Thanks @kristian-ziga)
• The "yes-always" options for prompts are now correctly respected in Aider modes.
• Ensured the selected edit format is correctly applied when working in Architect mode.
• Implemented a warning specifically for Gemini models regarding their tool limitation with search grounding.

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.31.1

Hotfix

• fixed 'undefined' error when migrating from older versions

Complete release notes for version 0.31.0 are available here.

0.31.0

Highlights

✨ Dynamic Model Loading & Enhanced Model Selection

Getting started with AiderDesk is now smoother than ever! The model selector for Agent mode now dynamically loads models based on the providers you have configured and for which API keys are provided. This means you'll instantly see available models without manual configuration, making the onboarding process much simpler for new users. We've also enhanced model selection for both Agent and Aider modes, ensuring that once you configure a provider, the appropriate model is automatically set, allowing you to jump straight into your tasks. While this significantly improves the initial experience, we're continuously working on further enhancements to model selection and onboarding. If your desired model isn't listed, you can still manually type its name with the provider's prefix (e.g., openai/model-007).

🛠️ Smarter Power Tool Messages UX

We've refined the user experience for Power Tool messages to make your agent's workflow clearer and less cluttered. Power Tool messages are now better recognizable with distinct icons and improved labeling. Furthermore, tool messages now feature smarter expanded and collapsed states based on their type and result. This ensures that when you're reviewing the agent's steps, distracting or less critical messages are automatically collapsed, providing a much cleaner and more focused view of the important actions and outcomes.

📂 Improved Monorepo Support

Working with complex repository structures is now more robust. We've added better support for opening projects located in a subfolder of a Git root, such as within a monorepo. This enhancement ensures AiderDesk correctly recognizes and manages your project context even in these advanced setups.

🚀 Performance Boost with React 19

AiderDesk has been updated to React 19, bringing potential performance improvements and laying the groundwork for future UI enhancements. This update helps ensure a snappier and more responsive user experience.

↩️ Enhanced Undo Last Commit

You now have more control over your Git history! The /undo command has been enhanced to provide better support for undoing the last commit, even outside of active Aider sessions. This means you can more reliably revert unpushed commits, giving you greater flexibility in your development workflow.

Other Changes

• Added None reasoning effort option for the OpenAI provider.
• Implemented support for an optional autoApprove property in custom commands.
• Properly loading messages from the last session on startup for continuity.
• Automatically setting configured provider models when API keys are provided.

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.30.0

Highlights

🌐 Powerful Remote Access & Security for AiderDesk Server

We're introducing robust controls for AiderDesk's remote capabilities. In Settings -> Server, you can now:

• Manually Start/Stop the Server: Take direct control over when your AiderDesk API and browser access are active.
• One-Click Cloudflare Tunnel: Instantly create a secure Cloudflare tunnel to access your AiderDesk instance via a public URL, making it accessible from anywhere.
• Basic Authentication: Secure your server access and API with configurable basic authentication (username and password), ensuring only authorized users can connect.

📱 Responsive Design for Mobile Devices

To provide a seamless experience when accessing AiderDesk remotely on mobile phones, the application's layout has been updated with a simple but functional responsive design. This ensures a better user interface and interaction on smaller screens.

✨ Enhanced Subagent Memory Modes

Your specialized Subagents just got more flexible! We've extended their context memory options to include "Full Context" or "Last Message". This means you can now configure your subagents to remember their entire conversation history or just the last messages, allowing for more precise control over their operational context and further optimizing token usage.

🔠 Customizable Font Size

Following up on our previous theme and font family enhancements, you can now change the font size within the application. This new setting offers even more personalization, allowing you to tailor AiderDesk's appearance to your visual preferences. Thanks @kristian-ziga.

Other Changes

• Added token information for context files, providing better insight into their contribution to the overall context. (Thanks @kristian-ziga)
• Fixed an issue where info message types were appearing prematurely before the end of diffs when using fast models.

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.29.0

Highlights

✨ Remote AiderDesk in Your Browser (Preview) & Full API Layer

AiderDesk is breaking free from your desktop! You can now access your running instance remotely via your web browser at http://localhost:24337. This means you can work on your coding tasks wherever you are, provided you have external access to your computer. This initial preview supports most functionalities, with some limitations for browser-specific interactions like file dialogs or terminal access.

For this preview, the remote server is enabled by default for testing. In upcoming releases, you can expect options to manage the server connection and even create a Cloudflare tunnel to connect from anywhere via an external link.

To support this, we've implemented a full API layer, opening the door for custom UI clients, including the possibility of mobile versions! Additionally, you can now run AiderDesk in a headless mode by setting the AIDER_DESK_HEADLESS=true environment variable, perfect for server environments or automated workflows.

🧠 Subagents with Context Memory

Your specialized Subagents just got smarter! We've added a "Has context memory" option, allowing your subagents to preserve their conversation context within the same session. This is incredibly useful for subagents performing research or online searches, as they can now remember previous interactions and avoid redoing work. A big thank you to @MotherSoraka for this valuable suggestion!

🚫 Granular Control Over Subagent Usage

Sometimes you want your main agent to work independently without delegating tasks to subagents. We've introduced a new "Can use subagents" configuration option in the Agent profile settings. This allows you to precisely control whether a specific agent profile is allowed to invoke and utilize your configured subagents.

Other Changes

• Context information is now updated correctly after Aider mode requests.
• A message is now shown when a commit message is being generated after modifications.
• Implemented an events manager layer with Socket.IO support for real-time updates, which underpins the new remote access.

💡 We’d love to hear your thoughts! Your feedback helps us improve AiderDesk - feel free to share any suggestions or issues. Happy coding! 🚀

0.28.1

Hotfix

• fixed the issue with semantic search Power tool not having the required binary and not working as expected

Complete release notes for version 0.28.0 are available here.

0.28.0

Highlights

✨ Introducing Subagents: Specialized AI Assistants for Enhanced Productivity

Unleash a new level of efficiency with Subagents! You can now create specialized AI assistants from any agent profile, each with its own customized system prompt, tool access, and settings. Subagents are designed to handle specific types of tasks, whether invoked automatically by the main agent or on-demand by you.

Key Benefits:

• Context Preservation: Subagents operate with a focused context, preventing pollution of your main conversation.
• Specialized Expertise: Configure subagents with detailed prompts for specific domains, leading to higher success rates.
• Cost Optimization: Strategically select cost-effective models (e.g., Gemini Flash, Claude Haiku) for routine subagent tasks like code reviews, and premium models for complex work.
• Flexible Permissions: Control tool access and approval settings for each subagent.

By offloading specific tasks, subagents help optimize your main context window and make your AI interactions significantly more cost-efficient. We believe this is a game-changing feature and are eager to see what specialized subagents you create! Expect continuous improvements to this functionality in future releases.

Detailed information with some examples can be found in our documentation.

✅ Confirm Before Edit for Aider Coder

Gain more control over AI-generated code changes! When using Aider in Code mode, you now have the option to confirm before edit. This allows you to review the modifications Aider proposes with a diff viewer before accepting them or providing feedback for further improvements to the model.

🌐 Built-in Web Scraping with fetch Power Tool

Your main agent now has the powerful capability to scrape websites it encounters during a conversation, thanks to the new fetch power tool. We've also removed the external Playwright dependency for web scraping. This means the /web command now performs its scraping internally with the same high quality, simplifying your setup and enhancing performance.

📂 Seamless Drag & Drop for Context Files

Managing your context files just got easier and more intuitive! You can now drag and drop files directly into the Context Files panel from external applications or within your project. Files dropped from outside the current project are added as read-only, while those from within the project are fully editable.

🧠 'Minimal' Reasoning Effort for Next-Gen Models

We've added a 'minimal' reasoning effort option for the OpenAI provider and the /reasoning-effort command when using Aider. This is the preferred setting when working with highly capable models like GPT-5, allowing them to operate more efficiently while maintaining high-quality outputs for coding tasks.

Other Changes

• Added a model selector to the Agent profile settings for more granular control.
• The selected model is now correctly used when compacting the conversation history.
• Improved the scrollbar style in the agent selector for better usability with many profiles.
• Fixed an issue where the application could get stuck when dropping files.
• Updated the temperature range for models to 0-2 with a 0.5 step for finer control.
• The file edit tool now gracefully handles scenarios with no changes, preventing unnecessary errors.
• Added a "scroll to bottom" button to the message area for improved navigation in long conversations.
• The AGENTS.md file is now included in the system prompt by default when using Agent mode, providing better context to the AI.
• Fixed an issue causing duplication in the thinking section of messages with some OpenRouter models.
• Context messages and files are now properly used with context-aware commands (e.g., /tokens).
• Relevant model selectors are now dynamically shown in the Project bar based on the current mode and available tools.