Aider Desk: Aider THE GOAT is MAKING A COMEBACK! This Aider-based GUI Coder is AWESOME!

Key Concepts

• Ader: An agentic coder, notable for its use of XML tool format for real-world tasks.
• Aderesk: A new open-source graphical interface built on top of Ader, offering enhanced features and ease of use.
• Electron App: A desktop application framework that Aderesk utilizes, embedding the Ader core.
• Gitaware Whole File Editing Engine: The core editing mechanism of Ader, preserved in Aderesk.
• Versel AI SDK: Used by Aderesk to build a custom AI agent for planning before code generation.
• Agent Mode: A mode in Aderesk that allows autonomous file reading, editing, and task execution.
• Code Mode: A mode in Aderesk where the AI only operates on manually added context files.
• Architect Mode: A mode that separates planning and implementation phases for changes.
• Model Endpoint: A system for connecting to various AI model providers, such as Open Code.
• Usage Dashboard: A feature in Aderesk to track API token usage and associated costs.

Aderesk: A Graphical Interface for Agentic Coding

This video introduces Aderesk, a new open-source project that enhances the capabilities of Ader, an established agentic coder. Aderesk transforms Ader into a user-friendly Electron app, providing a graphical interface and a suite of additional features while retaining Ader's core functionalities.

Core Functionality and Installation

Aderesk embeds the original Python Ader core, ensuring that all existing Ader flags and its "gitaware whole file editing engine" remain functional and unchanged. A significant advantage of Aderesk is its "one-click install" process, eliminating the need for manual Python, pip, or Rust toolchain installations. The application automatically handles the installation of all necessary libraries.

User Interface and Project Management

Upon launching Aderesk, users are prompted to configure their primary AI provider and can optionally set up sub-agents. The interface requires users to open a project folder to begin working.

• Project Tabs: Multiple projects can be opened and managed as tabs at the top of the interface, similar to a web browser.
• Task Delegation: A left-hand panel displays delegated tasks within a project, facilitating the management of multiple agents working concurrently on a single project.

Model and Tool Configuration

Aderesk offers granular control over AI model selection and tool usage:

• Model Configuration: Users can select a "main model" for primary operations and a "weak model" for less demanding tasks like generating commit messages or summarizing conversation history. Supported models include options like "flash" and "haiku."
• Edit Tool Format: The format for editing tools can be chosen from options such as "diff," "diff fenced," "whole," and "udiff." "Diff" is the default setting.

Interaction Modes

Aderesk provides several distinct interaction modes tailored for different coding workflows:

• Code Mode: This is the default mode, designed for general coding tasks. It strictly operates on files manually added to the "context panel" on the right, preventing it from accessing other files in the project.
• Agent Mode: This mode is recommended for autonomous operations, similar to Claude Code. It allows the AI to read and edit files, and perform tasks autonomously. Context is dynamically added based on user prompts. This mode also enables the configuration of different agent types, specifying enabled tools, MCPs (presumably "Model Configuration Parameters"), rules, and context.
  • Power Tools: Included by default, offering all edit-related functionalities.
  • Ader with Power Search: Enhances file searching and provides advanced capabilities.
• Architect Mode: This mode separates the planning and implementation stages. A "planning model" first outlines the necessary changes, which are then handed off to an "editor model" for execution.
• Context Mode: Allows users to add specific files or entire folders to the context panel for the AI to work with.
• Terminal Option: Provides an integrated terminal for direct command-line interaction.

Context Management and Settings

• Context Panel: Located on the right, this panel allows users to explicitly add files or folders for the AI to consider, particularly useful in "Code Mode."
• Settings: Accessible from the top right, this section allows customization of fonts, colors, and other interface elements.
• Model Library: Users can configure their preferred AI providers through this option, leveraging Open Code's model endpoint to connect with a wide range of providers.

Usage Tracking and Performance

• Usage Dashboard: A valuable feature that tracks API token consumption and the costs incurred for various tasks and requests.
• Performance: The interface is described as "snappy," with tasks executing quickly and smoothly. The AI's responsiveness is highlighted.

Advanced Features and Capabilities

• Edit Task Implementation: When given an edit task, Aderesk presents an implementation plan with a to-do list. Users can manually check off tasks, or the AI can mark them as completed as it progresses.
• Intuitive Interface and Agent: The interface is deemed "extremely intuitive," and the AI agent is praised for its effectiveness, particularly for professional use cases involving multiple agents on diverse tasks. The combination of Ader's agentic capabilities with Aderesk's interface is considered "really powerful."

Conclusion

Aderesk is presented as a significant advancement for Ader, offering a user-friendly graphical interface, simplified installation, and enhanced control over AI agent workflows. While the full extent of its capabilities has not been exhaustively tested, the results are reported to be on par with leading AI coding assistants like Claude and Open Code. The project is encouraged for adoption by users seeking a more accessible and powerful agentic coding tool.