This plugin provides AI-powered code assistance for Embarcadero RAD Studio IDEs, supporting both online services (ChatGPT, WriteSonic, YouChat) and offline Ollama models. It aims to boost developer productivity by offering features like inline code generation, bug finding, test creation, and code explanation directly within the IDE.

How It Works

The plugin integrates with various AI backends via their respective APIs. For online services, it sends prompts and receives responses through standard API calls, requiring API keys. For offline use, it connects to a locally running Ollama server, allowing developers to leverage local AI models without internet connectivity or API costs. This dual approach offers flexibility and cost-efficiency.

Quick Start & Requirements

• Installation: Via Getit package manager, Delphinus package manager, or direct build/install from source in Delphi.
• IDE Support: Delphi 10.1 Berlin and later (including 11.x Alexandria, 12.x Athens). A limited version is available for XE2-XE8.
• Note: The plugin is not supported for RAD Studio 12.2 and above due to the introduction of a built-in AI engine.
• Dependencies: XSuperObject library (included). For offline use: Ollama and at least one downloaded model (e.g., codellama).
• API Keys: Required for online services.

Highlighted Details

• Supports online AI services (ChatGPT, WriteSonic, YouChat) and offline Ollama.
• Offers inline code assistance, context menu actions (bug finding, optimization, comments), and class-specific AI operations.
• Includes a history feature with fuzzy string matching for saved tokens.
• Provides an option to use a proxy server for AI requests.

Maintenance & Community

The project is maintained for RAD Studio versions up to 12.1. The README encourages stars, PRs, discussions, and issues, with a note to check closed issues first.

Licensing & Compatibility

The project is made with Delphi. No specific license is mentioned in the README, but it relies on third-party AI services which have their own terms and potential costs. Compatibility is explicitly stated for Delphi versions 10.1+ and limited support for XE2-XE8.

Limitations & Caveats

The plugin is not compatible with RAD Studio 12.2+. Offline mode performance is dependent on user hardware and model size, and users are cautioned against opening issues related to speed. Large classes may exceed API limitations for context-aware operations.