PaletteBrain

Provides a native macOS menu bar application that intercepts keyboard shortcuts or menu interactions to spawn ChatGPT chat windows from any application context without requiring browser navigation. Implements a global hotkey listener (likely using macOS Accessibility APIs or Carbon Event Manager) that captures user input and routes it to an embedded or proxied ChatGPT interface, maintaining session state across application switches.

Enables users to select code snippets or entire files from their editor and submit them to ChatGPT with a single action, likely via clipboard monitoring or direct file path integration. The implementation probably uses macOS pasteboard APIs to detect code selection and automatically format it with language hints (e.g., markdown code blocks with language tags) before sending to ChatGPT, preserving syntax highlighting context.

Maintains a conversation thread that persists across application switches and menu bar interactions, allowing users to reference previous messages and build multi-turn conversations without losing context. Likely implemented via local SQLite or JSON file storage of conversation metadata (message IDs, timestamps, content) synced with ChatGPT's session token, enabling users to resume conversations even after closing the menu bar app.

Allows users to select which ChatGPT model version (GPT-4, GPT-3.5, etc.) to use for queries and configure system-level settings like temperature, max tokens, or API endpoint. Implemented via a preferences pane or settings modal that stores configuration in macOS UserDefaults or a local config file, then passes these parameters to ChatGPT API calls or web session initialization.

Provides pre-built prompt templates or macros for common tasks (code review, documentation generation, debugging) that users can trigger via keyboard shortcuts or menu selections. Implemented as a template library stored locally (JSON or plist format) with variable substitution (e.g., {{selected_code}}, {{file_name}}) that gets expanded at runtime and sent to ChatGPT.

Automatically formats ChatGPT responses with markdown rendering, syntax highlighting for code blocks, and copyable code snippets. Likely uses a markdown parser (e.g., CommonMark or a lightweight alternative) to convert ChatGPT's markdown output into formatted text/HTML, with native macOS text rendering for proper typography and code block styling.

Attempts to detect the active application and file type to provide contextual suggestions or auto-format prompts. Likely uses macOS Accessibility APIs to query the frontmost application and file metadata (via NSWorkspace or similar), then adjusts ChatGPT prompts or response formatting based on detected context (e.g., Python code in VS Code vs. Markdown in Notion).

Allows users to define custom keyboard shortcuts for triggering ChatGPT access, submitting prompts, or executing prompt templates. Implemented via macOS event monitoring (likely using Carbon Event Manager or newer Cocoa APIs) to register global hotkeys that work across all applications, with conflict detection and customization via preferences UI.

Provides intelligent copy functionality that preserves formatting (syntax highlighting, markdown structure) when copying ChatGPT responses to clipboard, with options to copy as plain text, markdown, or formatted code. Likely uses macOS pasteboard APIs with multiple data types (NSString, NSAttributedString, NSData) to support rich text copying across different applications.