KrrishCoder - Context-Aware Code Retrieval

View on VS Code Marketplace

KrrishCoder is a VS Code extension that helps developers find relevant code using natural-language, retrieval-augmented search.

Features

• Ask natural-language questions like "Where is my login code?"
• Search across the whole opened Python workspace, not only one file
• Retrieve the top 5 relevant functions/classes
• Show file path and line range for each result
• Click a result to open the file directly at the matching line
• Use a chat-style UI for asking codebase questions
• Cache embeddings in .cache_rag for faster repeated searches
• Re-chunk and re-embed only changed Python files using fast mtime/size comparison
• Use fastembed and numpy; PyTorch and sentence-transformers are not required

Usage

1. Install the extension.
2. Open a Python project folder in VS Code.
3. Open the Command Palette with Ctrl+Shift+P or Cmd+Shift+P.
4. Run RAG Chat: Ask code questions.
5. Type your question and click a result to navigate to the code.

You can also use:

• RAG Search: Context-Aware Code Search for quick-pick search results
• RAG: Rebuild Index to force a fresh workspace index

On activation, the extension checks Python dependencies and installs the packages from requirements.txt with pip --user if they are missing.

Current Language Support

• Python

Future Plans

• Support for Java, C++, and JavaScript
• Improved ranking of code snippets
• Integration with AI assistants
• More detailed code explanations inside chat

Release Notes

2.0.0

• Added workspace-wide Python code retrieval
• Added chat-style VS Code webview UI
• Added clickable file and line-range results
• Replaced sentence-transformers with lightweight fastembed
• Removed PyTorch requirement
• Added automatic Python dependency installation
• Added incremental cache refresh for changed files only
• Modularized the Python retrieval code

1.0.0

• Initial release with Python support