feat: Enhance context gathering for AI suggestions with project-wide awareness

Author: TheUnknown550

README.md

@@ -1,7 +1,10 @@
 # Local AI Copilot for VS Code
 
+
 A VS Code extension that provides GitHub Copilot-like functionality using locally running Ollama models.
 
+Now with **smart project-wide context**: The extension automatically gathers code and documentation from your workspace for every AI request, making suggestions and edits much more intelligent and Copilot-like.
+
 ## Features
 
 - 🤖 **Chat Interface** - Interactive AI chat panel similar to GitHub Copilot
@@ -12,6 +15,10 @@ A VS Code extension that provides GitHub Copilot-like functionality using locall
 - 🎨 **Beautiful UI** - Modern interface that matches VS Code's theme
 - 🚀 **Multiple Models** - Choose between different Ollama models
 
+- 🧠 **Project-Wide Context Awareness** – AI suggestions and edits use key files from your workspace (README, SETUP, main, server, extension, `.py`, `.ts`, `.js`, `.md`).
+- 🔍 **Intelligent Suggestions & Edits** – AI can reason about your whole project, not just the current file or cursor.
+- 🛠️ **Smart Code Actions** – Insert, replace, and preview code with context-aware suggestions and diff previews.
+
 ## Prerequisites
 
 1. **Ollama** - Install from [https://ollama.ai](https://ollama.ai)
@@ -59,6 +66,14 @@ Then press `F5` to launch the extension in a new VS Code window.
 
 ### Chat Interface
 
+## Smart Project Context
+
+- The extension automatically gathers context from up to 5 key files in your workspace (README, SETUP, main, server, extension, `.py`, `.ts`, `.js`, `.md`) for every AI request.
+- It combines this with the current editor's context and selection.
+- The AI will now be able to reason about your whole project, not just the current file or cursor.
+
+This makes code suggestions, edits, and completions much smarter and more Copilot-like!
+
 1. Open Command Palette (`Ctrl+Shift+P` or `Cmd+Shift+P`)
 2. Type "Local AI Copilot: Chat"
 3. Start chatting with your AI!
@@ -89,6 +104,13 @@ Choose your preferred model from the dropdown in the chat panel:
 2. **Python Server**: Handles communication with Ollama
 3. **Ollama**: Runs the AI models locally on your machine
 
+**Advanced:**
+
+When you ask for code generation, editing, or explanation, the extension:
+1. Collects relevant code and documentation from your workspace.
+2. Combines it with your current selection or file context.
+3. Sends this rich context to the AI backend for smarter, project-aware results.
+
 ```
 VS Code Extension (TypeScript)
        ↓

vscode-extension/src/extension.ts

@@ -169,7 +169,25 @@ async function handleChatMessage(message: any) {
         let context = '';
         let taskType = 'chat';
 
-        // Smart context gathering
+        // Project-wide smart context gathering
+        const workspaceFolders = vscode.workspace.workspaceFolders;
+        let projectContext = '';
+        if (workspaceFolders && workspaceFolders.length > 0) {
+            const folder = workspaceFolders[0].uri.fsPath;
+            // Gather up to 5 key files (README, setup, main, server, extension, etc.)
+            const keyFiles = fs.readdirSync(folder)
+                .filter(f => /README|SETUP|main|server|extension|\.py$|\.ts$|\.js$|\.md$/i.test(f))
+                .slice(0, 5);
+            for (const file of keyFiles) {
+                try {
+                    const filePath = path.join(folder, file);
+                    const fileContent = fs.readFileSync(filePath, 'utf8');
+                    projectContext += `\n\n--- File: ${file} ---\n` + fileContent.slice(0, 2000); // Limit size per file
+                } catch (e) { /* skip unreadable files */ }
+            }
+        }
+
+        // Smart context from current editor
         if (editor) {
             const selection = editor.selection;
             const selectedText = editor.document.getText(selection);
@@ -189,6 +207,11 @@ async function handleChatMessage(message: any) {
             }
         }
 
+        // Combine project context and editor context
+        if (projectContext) {
+            context = (context ? context : '') + '\n\n--- Project Context ---\n' + projectContext;
+        }
+
         const response = await callAI(message.prompt, taskType, context, message.model);
         console.log('AI response from server:', response);