vscode-chat-gpt

Provides a dedicated sidebar chat panel that maintains a conversational interface with OpenAI's gpt-3.5-turbo model via streaming API calls. The extension sends user queries directly to OpenAI's chat completions endpoint with configurable temperature (default 0.7) and renders responses incrementally as they arrive, reducing perceived latency. Chat history is maintained in-session within the sidebar panel, with the last 10 queries persisted in VSCode extension state for quick reference.

Analyzes selected code in the editor and generates inline comments using OpenAI's text-davinci-003 model with fixed parameters (temperature 0.0, top_p 0.1, max_tokens 2048). The extension captures the selected text via VSCode's editor API, sends it to the completions endpoint with an implicit 'add comments' prompt, and inserts the generated comments back into the editor at the selection location. Works across all programming languages supported by VSCode.

Maintains a history of the last 10 user queries in a dedicated 'Query History' view within the sidebar panel (added in v1.0.0). The extension stores queries in VSCode's extension state API, which persists data across editor sessions. Users can click on a previous query to re-execute it or view the original prompt. The history is limited to 10 items to prevent excessive state bloat, and older queries are automatically discarded when the limit is exceeded.

Streams responses from OpenAI's chat completions API and renders them incrementally in the sidebar chat panel as tokens arrive, rather than waiting for the complete response. The extension uses OpenAI's streaming API (stream=true by default) and updates the UI with each token chunk, creating a real-time typing effect. This reduces perceived latency and allows users to start reading responses before generation completes. Streaming is enabled by default with no documented toggle option.

Generates docstrings and API documentation for selected code using OpenAI's text-davinci-003 model, but restricts this capability to JavaScript, TypeScript, Java, and C# due to model training specificity. The extension detects the file extension or language mode, validates against the supported language list, and only enables the 'Add Documentations' context menu command if the current file matches. Generated documentation is inserted at the selection location with fixed parameters (temperature 0.0, max_tokens 2048).

Analyzes selected code and generates refactoring suggestions using text-davinci-003 with deterministic parameters (temperature 0.0, top_p 0.1, max_tokens 2048). Like documentation generation, this capability is restricted to JavaScript, TypeScript, Java, and C# to ensure model quality. The extension validates the file language before enabling the 'Refactor' context menu command, sends the selected code to the completions endpoint with an implicit refactoring prompt, and returns suggestions as text output without automatic code replacement.

Generates images from natural language text prompts using OpenAI's DALL-E API integrated into a dedicated 'Image Generation' tab in the sidebar panel (added in v1.2.0). The extension sends user prompts to the DALL-E endpoint with fixed parameters (size 1024x1024, n=1 for single image per request) and displays the generated image URL in the sidebar. Users can view, copy, or download generated images directly from the extension UI.

Adds a clickable icon to VSCode's Activity Bar (left sidebar) that toggles the extension's main chat and image generation panel on/off. This provides a single-click entry point to the extension's functionality without requiring command palette invocation or keyboard shortcuts. The Activity Bar icon was added in v0.0.2 and serves as the primary UI affordance for launching the extension's sidebar panel.

Stores the OpenAI API key securely using VSCode's built-in secrets storage mechanism (implied by 'safely stored' documentation). The extension prompts users to enter their API key via the Settings panel (added in v1.2.0) and persists it in the workspace configuration. The key is retrieved on each API call and used to authenticate requests to OpenAI endpoints. VSCode's secrets API encrypts sensitive data at rest and prevents exposure in version control or plain-text configuration files.

Exposes the temperature parameter for chat completions (gpt-3.5-turbo) with a default value of 0.7, allowing users to adjust response creativity and determinism via the Settings panel (added in v1.2.0). Temperature controls the randomness of token selection: lower values (closer to 0.0) produce more deterministic, focused responses, while higher values (up to 2.0) increase diversity and creativity. The setting is persisted in VSCode workspace configuration and applied to all subsequent chat API calls.

Allows users to configure the output size for DALL-E image generation via the Settings panel (added in v1.2.0), with a default of 1024x1024 pixels. The setting is persisted in workspace configuration and applied to all subsequent image generation requests. Users can select from DALL-E's supported sizes (typically 256x256, 512x512, 1024x1024) to balance image quality, generation time, and API cost.

Integrates three code-focused commands (Add Comments, Add Documentations, Add Refactoring) into VSCode's right-click context menu with platform-specific keyboard shortcuts. Windows/Linux use Ctrl+Alt+[C/D/R], while macOS uses Shift+Cmd+[C/D/R]. The extension registers these commands via VSCode's command API and binds them to keybindings in the extension's package.json. Commands are conditionally enabled based on file language (documentation and refactoring restricted to JS/TS/Java/C#) and only appear in the context menu when a code selection exists.