Gemini Assistant

Analyzes user-selected code snippets by capturing the current editor selection and sending it to Google's Gemini API via authenticated REST calls, returning markdown-formatted analysis rendered in a dedicated sidebar panel. The extension integrates with VS Code's context menu to trigger analysis without requiring manual copy-paste, maintaining the selection state and file context during the API round-trip.

Extends selection-based analysis to entire file contents by reading the active editor's full buffer and submitting it to Gemini for comprehensive analysis. The extension handles file-level context by capturing the complete source code and sending it as a single API request, enabling broader pattern recognition and architectural feedback compared to snippet-level analysis.

Enables developers to ask natural language questions about code by composing queries in the sidebar panel and receiving Gemini-generated responses. The extension maintains a conversation history within the sidebar, allowing follow-up questions that reference previous context, with responses rendered as markdown in the panel. Each query is sent to Gemini with the current editor context (selected code or file, depending on user action).

Provides a dropdown configuration interface in VS Code Settings to select from six pre-configured Google Gemini models (gemini-2.5-pro-exp-03-25, gemma-3-27b-it, gemini-2.0-flash, gemini-2.0-flash-lite, gemini-pro) plus a 'Custom' option that allows users to specify arbitrary model names. The extension routes all API requests through the selected model, enabling developers to trade off cost, latency, and capability without code changes.

Implements authentication to Google's Gemini API by storing an API key in VS Code's settings system (via the 'Gemini Assistant: Api Key' configuration field). The extension reads this key on startup and includes it in all API requests to authenticate with Google's servers. The key is stored in VS Code's local settings file, with encryption status unknown.

Formats all Gemini API responses as markdown and renders them in a dedicated sidebar panel with full markdown support (headers, code blocks, lists, links, etc.). The extension parses the API response text and applies markdown rendering rules, displaying formatted output in the panel UI rather than raw text. Code blocks within responses are syntax-highlighted based on language hints.

Captures the entire conversation history from the sidebar panel and exports it as a static HTML file that can be saved to disk. The export includes all user queries and Gemini responses in chronological order, preserving markdown formatting and code blocks. The exported HTML file is self-contained and can be opened in any web browser for review or sharing.

Provides a dedicated sidebar panel in VS Code that displays Gemini responses, maintains conversation history, and serves as the primary UI for interacting with the extension. The panel persists across file switches and editor actions, allowing developers to reference previous responses while working on code. The panel includes controls for triggering analysis, composing queries, and exporting history.

Integrates with VS Code's right-click context menu to trigger code analysis on selected text without requiring command palette or keyboard shortcuts. When a user right-clicks on selected code, the extension adds menu items that send the selection to Gemini for analysis or questioning. The context menu integration is automatically registered when the extension loads.

Handles the low-level mechanics of sending code and queries to Google's Gemini API, including request formatting, authentication header injection, and response parsing. The extension constructs API requests with the selected model, API key, and user input, sends them over HTTPS to Google's endpoints, and parses JSON responses. Error handling for API failures is not documented.