GPT Code

Generates code snippets and complete functions by accepting natural language descriptions through a VS Code sidebar interface, sending prompts to OpenAI's GPT models (3.5-turbo or GPT-4 with whitelisting), and inserting generated code directly into the active editor. The extension maintains conversation history within the session to allow iterative refinement of generated code through follow-up prompts.

Provides language-aware code completion suggestions by analyzing the current file's language context and sending partial code or cursor position to OpenAI, returning contextually appropriate completions. The extension claims support for multiple programming languages through language detection and language-specific prompt engineering, though specific supported languages are not enumerated.

Exposes extension settings and configuration through VS Code's command palette via the 'GPT Code Configure' command, allowing users to set API keys, select models, configure proxy endpoints, and adjust sentiment/mode settings without manually editing configuration files. Configuration is stored in VS Code's extension settings storage.

Analyzes selected code blocks or entire files and generates human-readable explanations by sending code to OpenAI, returning detailed descriptions of functionality, logic flow, and purpose. The explanation is displayed in the sidebar chat interface, allowing developers to ask follow-up questions about specific code sections through the conversation history mechanism.

Accepts natural language refactoring instructions (e.g., 'extract this function', 'rename variables for clarity', 'convert to async/await') and applies transformations to selected code by sending the code and instruction to OpenAI, then inserting the refactored result back into the editor. The extension supports editing of previously generated responses through a 'Historic message edit' feature, allowing users to regenerate or modify refactoring results without re-selecting code.

Generates responses to code-related questions with configurable sentiment or tone (feature listed but specific sentiment options and implementation details are undocumented). The extension likely applies prompt engineering or post-processing to adjust the emotional tone or formality of responses based on user configuration, though the exact mechanism and available sentiment modes are unknown.

Supports multiple operational modes (feature listed but specific modes are not documented) that likely adjust how the extension processes prompts, accesses context, or generates responses. Modes may include variations such as 'quick mode' for fast suggestions, 'detailed mode' for comprehensive explanations, or 'code-focused mode' for generation-heavy tasks, though the exact modes and their effects are unknown.

Supports configuration of proxy API endpoints to route OpenAI requests through alternative servers, enabling access in regions where OpenAI's API is blocked or restricted. The extension accepts custom proxy endpoint configuration in settings, allowing users to specify alternative API gateways or regional mirrors that forward requests to OpenAI's infrastructure.

Allows configuration of OpenAI model selection between GPT-3.5-turbo and GPT-4, with GPT-4 access requiring explicit whitelisting from OpenAI (a non-standard requirement suggesting custom implementation or outdated documentation). The extension sends requests to the selected model via OpenAI's API, with model choice affecting response quality, latency, and cost.

Maintains conversation history throughout a VS Code session, allowing users to reference previous prompts, responses, and generated code in follow-up questions without re-selecting code or losing context. The conversation is displayed in the sidebar chat interface and supports editing of previous messages through the 'Historic message edit' feature, enabling users to regenerate responses or refine earlier prompts.

Provides a dedicated sidebar panel within VS Code for chat-based interaction with the AI, displaying conversation history, allowing natural language input, and showing AI responses alongside code editor context. The sidebar interface integrates with the editor's selection and file context, enabling users to reference code without explicit copying or pasting.