Safurai - AI Assistant for Javascript, Python, Typescript & more

Provides intelligent code suggestions across 15+ programming languages (JavaScript, Python, TypeScript, Java, C++, C#, Go, Rust, PHP, Kotlin, etc.) by analyzing the current file context and cursor position. Uses LLM-based completion that understands syntax and semantic patterns within the editor buffer, integrating directly with VS Code's IntelliSense API to surface suggestions inline without context switching.

Analyzes selected code blocks or entire functions and generates human-readable explanations of what the code does, how it works, and why certain patterns are used. Integrates with VS Code's command palette and context menus to allow one-click explanation generation, then displays results in a side panel or inline hover. Supports generating documentation in multiple formats (docstrings, JSDoc, Javadoc, etc.) based on language context.

Identifies code sections that can be refactored for readability, performance, or maintainability by analyzing syntax patterns, variable naming, and structural inefficiencies. Provides refactoring suggestions (extract function, rename variable, simplify logic, remove duplication) with before/after diffs. Uses LLM reasoning to understand intent and propose semantically equivalent but improved code, with one-click application of changes directly to the editor buffer.

Scans code for potential bugs, logic errors, and anti-patterns by leveraging LLM reasoning over syntax and semantic analysis. Identifies issues like null pointer dereferences, off-by-one errors, type mismatches, and common pitfalls in the selected language. Provides explanations of why the code is buggy and suggests fixes with reasoning, allowing developers to understand the issue before applying the fix.

Analyzes code for performance bottlenecks, inefficient algorithms, and resource usage patterns. Suggests optimizations such as algorithmic improvements, caching strategies, lazy loading, and language-specific performance best practices. Provides before/after performance impact estimates and explanations of optimization trade-offs (e.g., memory vs. speed). Integrates with the editor to highlight optimization opportunities and apply changes incrementally.

Enables semantic search across the codebase using natural language queries (e.g., 'find functions that handle user authentication'). Uses LLM embeddings or semantic understanding to match code intent rather than keyword matching. Integrates with VS Code's search UI to display results with context snippets, allowing developers to navigate to relevant code without knowing exact function names or file locations.

Provides an interactive chat interface within VS Code where developers can ask questions about code, request explanations, or get suggestions. The chat maintains context of the currently selected code or open file, allowing questions like 'how does this function work?' or 'what's a better way to write this?'. Uses multi-turn conversation to refine questions and provide iterative assistance, with the ability to apply suggested code changes directly from chat responses.

Allows developers to choose and switch between multiple LLM backends (ChatGPT, Bard, GPT-4, and potentially others) without changing workflows or re-configuring the extension. Provides a settings UI to select the preferred backend and manage API keys. Enables A/B testing different models or using cost-optimized backends for different tasks (e.g., GPT-3.5 for simple completions, GPT-4 for complex reasoning).

Integrates all Safurai capabilities into VS Code's command palette and allows custom keyboard shortcuts for frequently used actions (e.g., Ctrl+Shift+E for 'Explain Code', Ctrl+Shift+R for 'Refactor'). Provides a keybindings configuration UI where developers can map custom shortcuts to any Safurai command, enabling power-user workflows without mouse interaction.

Generates code snippets, boilerplate, and templates tailored to the specific language and framework context. Uses language-specific syntax rules and conventions (e.g., Python indentation, JavaScript async/await patterns, Java class structures) to ensure generated code is syntactically correct and idiomatic. Supports generating common patterns like API endpoints, database queries, test cases, and configuration files with language-appropriate syntax.