ai-first-cli 1.1.5 → 1.2.0

package/examples/README.md

@@ -40,7 +40,7 @@ touch examples/0X-your-framework.md
 git clone <repo>
 cd <repo>
 npx ai-first init
-# Read ai/ai_context.md
+# Read ai-context/ai_context.md
 # Now you understand the project in 2 minutes
 ```
 
@@ -50,13 +50,13 @@ npx ai-first init
 npx ai-first init
 
 # Then ask:
-"Read ai/ai_context.md and help me add feature X"
+"Read ai-context/ai_context.md and help me add feature X"
 ```
 
 ### For Technical Leads
 ```
 # Onboard new team members
 npx ai-first init
-# New hire reads ai/ai_context.md
+# New hire reads ai-context/ai_context.md
 # Instant project understanding
 ```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-first-cli",
-  "version": "1.1.5",
+  "version": "1.2.0",
   "description": "CLI tool that prepares any repository to be used effectively by AI coding agents",
   "main": "dist/index.js",
   "type": "module",

package/src/analyzers/symbols.ts

@@ -66,20 +66,24 @@ function parseFileForSymbols(file: FileInfo): Symbol[] {
     const content = readFile(file.path);
     const lines = content.split("\n");
     
-    if (file.extension === "ts" || file.extension === "tsx" || file.extension === "js" || file.extension === "jsx") {
+    if (file.extension === ".ts" || file.extension === ".tsx" || file.extension === ".js" || file.extension === ".jsx") {
       parseJavaScriptTypeScript(file, content, lines, symbols);
-    } else if (file.extension === "py") {
+    } else if (file.extension === ".py") {
       parsePython(file, content, lines, symbols);
-    } else if (file.extension === "go") {
+    } else if (file.extension === ".go") {
       parseGo(file, content, lines, symbols);
-    } else if (file.extension === "rs") {
+    } else if (file.extension === ".rs") {
       parseRust(file, content, lines, symbols);
-    } else if (file.extension === "java") {
+    } else if (file.extension === ".java") {
       parseJava(file, content, lines, symbols);
-    } else if (file.extension === "cs") {
+    } else if (file.extension === ".cs") {
       parseCSharp(file, content, lines, symbols);
-    } else if (file.extension === "cls" || file.extension === "trigger") {
+    } else if (file.extension === ".cls" || file.extension === ".trigger") {
       parseApex(file, content, lines, symbols);
+    } else if (file.extension === ".php") {
+      parsePHP(file, content, lines, symbols);
+    } else if (file.extension === ".rb") {
+      parseRuby(file, content, lines, symbols);
     }
   } catch {}
 
@@ -311,6 +315,139 @@ function parseJava(file: FileInfo, content: string, lines: string[], symbols: Sy
   }
 }
 
+/**
+ * Parse PHP files
+ */
+function parsePHP(file: FileInfo, content: string, lines: string[], symbols: Symbol[]): void {
+  for (let i = 0; i < lines.length; i++) {
+    const line = lines[i].trim();
+    
+    // Classes: class ClassName, public class ClassName
+    const classMatch = line.match(/^(?:public\s+|private\s+|protected\s+)?class\s+(\w+)/);
+    if (classMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, classMatch[1]),
+        name: classMatch[1],
+        type: "class",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    // Interfaces: interface InterfaceName
+    const interfaceMatch = line.match(/^(?:public\s+)?interface\s+(\w+)/);
+    if (interfaceMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, interfaceMatch[1]),
+        name: interfaceMatch[1],
+        type: "interface",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    // Functions: function functionName
+    const funcMatch = line.match(/^(?:public\s+|private\s+|protected\s+)?function\s+(\w+)/);
+    if (funcMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, funcMatch[1]),
+        name: funcMatch[1],
+        type: "function",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    // Constants: const CONST_NAME
+    const constMatch = line.match(/^const\s+(\w+)/);
+    if (constMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, constMatch[1]),
+        name: constMatch[1],
+        type: "const",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+  }
+}
+
+/**
+ * Parse Ruby files
+ */
+function parseRuby(file: FileInfo, content: string, lines: string[], symbols: Symbol[]): void {
+  for (let i = 0; i < lines.length; i++) {
+    const line = lines[i].trim();
+    
+    // Classes: class ClassName
+    const classMatch = line.match(/^class\s+(\w+)/);
+    if (classMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, classMatch[1]),
+        name: classMatch[1],
+        type: "class",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    // Modules: module ModuleName
+    const moduleMatch = line.match(/^module\s+(\w+)/);
+    if (moduleMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, moduleMatch[1]),
+        name: moduleMatch[1],
+        type: "module",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    const methodMatch = line.match(/^def\s+(\w+)/);
+    if (methodMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, methodMatch[1]),
+        name: methodMatch[1],
+        type: "function",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    const classMethodMatch = line.match(/^def\s+self\.(\w+)/);
+    if (classMethodMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, classMethodMatch[1]),
+        name: classMethodMatch[1],
+        type: "function",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+
+    // Constants: CONST_NAME =
+    const constMatch = line.match(/^([A-Z][A-Z0-9_]*)\s*=/);
+    if (constMatch) {
+      symbols.push({
+        id: generateSymbolId(file.relativePath, constMatch[1]),
+        name: constMatch[1],
+        type: "const",
+        file: file.relativePath,
+        line: i + 1,
+        export: true,
+      });
+    }
+  }
+}
+
 /**
  * Parse Apex (Salesforce) files
  */

package/src/commands/ai-first.ts

@@ -18,7 +18,7 @@ import { generateAIRules, generateAIRulesFile } from "../analyzers/aiRules.js";
 import { generateModuleGraph } from "../core/moduleGraph.js";
 import { loadIndexState, computeFileHash, getFilesToIndex, getChangedFilesGit } from "../core/indexState.js";
 import { chunkFiles } from "../core/chunker.js";
-import { generateEmbeddings, saveEmbeddings } from "../core/embeddings.js";
+import { generateEmbeddings, saveEmbeddings, createEmbeddingsTable } from "../core/embeddings.js";
 import { generateContextModules, createCCP, listCCPs, getCCP } from "../core/ccp.js";
 import { generateSemanticContexts } from "../core/semanticContexts.js";
 import { doctorMain } from "./doctor.js";
@@ -28,6 +28,7 @@ import { detectGitRepository, generateGitContext, analyzeGitActivity, getRecentF
 import { buildKnowledgeGraph, loadKnowledgeGraph } from "../core/knowledgeGraphBuilder.js";
 import { runIncrementalUpdate, detectChangedFiles } from "../core/incrementalAnalyzer.js";
 import { generateAllSchema } from "../core/schema.js";
+import { Database } from "sql.js";
 import ora from "ora";
 import process from "process";
 
@@ -53,7 +54,7 @@ export interface AIFirstResult {
 export async function runAIFirst(options: AIFirstOptions = {}): Promise<AIFirstResult> {
   const {
     rootDir = process.cwd(),
-    outputDir = path.join(rootDir, "ai"),
+    outputDir = path.join(rootDir, "ai-context"),
   } = options;
 
   const filesCreated: string[] = [];
@@ -419,7 +420,12 @@ Example queries (for AI agents):
       outputPath = path.join(rootDir, "ai", "index.db");
     }
 
-    const aiDir = path.join(rootDir, "ai");
+    const aiDir = path.join(rootDir, "ai-context");
+    
+    // Create output directory if it doesn't exist
+    if (!fs.existsSync(aiDir)) {
+      fs.mkdirSync(aiDir, { recursive: true });
+    }
 
     // Load existing index state for incremental indexing
     const existingState = loadIndexState(aiDir);
@@ -516,7 +522,23 @@ Example queries (for AI agents):
             console.log(`   Created ${chunks.length} chunks`);
             
             const { embeddings, model } = await generateEmbeddings(chunks);
-            saveEmbeddings(embeddings, aiDir, model, 384);
+            
+            const initSqlJs = (await import("sql.js")).default;
+            const SQL = await initSqlJs();
+            const dbPath = outputPath || path.join(rootDir, "ai", "index.db");
+            let db: Database;
+            if (fs.existsSync(dbPath)) {
+              const fileBuffer = fs.readFileSync(dbPath);
+              db = new SQL.Database(fileBuffer);
+            } else {
+              db = new SQL.Database();
+            }
+            createEmbeddingsTable(db);
+            saveEmbeddings(db, embeddings, model, 384);
+            const data = db.export();
+            const buffer = Buffer.from(data);
+            fs.writeFileSync(dbPath, buffer);
+            db.close();
             console.log("   ✅ Semantic indexing complete");
           } catch (error) {
             console.log("   ⚠️  Semantic indexing failed:", error);
@@ -614,7 +636,7 @@ Features:
     // Context command - generate AI context files OR symbol-specific context packet
     args.shift();
     let rootDir = process.cwd();
-    let outputDir = path.join(rootDir, "ai");
+    let outputDir = path.join(rootDir, "ai-context");
     let symbolArg: string | undefined;
     let depth = 1;
     let maxSymbols = 50;
@@ -628,7 +650,7 @@ Features:
           case "--root":
           case "-r":
             rootDir = args[++i];
-            outputDir = path.join(rootDir, "ai");
+            outputDir = path.join(rootDir, "ai-context");
             break;
           case "--output":
           case "-o":
@@ -1098,7 +1120,7 @@ Options:
     
     console.log("\n🗺️  Generating repository map...\n");
     const scan = scanRepo(rootDir);
-    const aiDir = path.join(rootDir, "ai");
+    const aiDir = path.join(rootDir, "ai-context");
     
     // files.json
     const filesJson = { files: scan.files.map(f => ({ path: f.relativePath, name: f.name, ext: f.extension })) };
@@ -1119,8 +1141,8 @@ Options:
     
     // repo-map.json (use local generateRepoMapJson function)
     const repoMapData = JSON.parse(generateRepoMapJson(scan.files.map(f => ({ relativePath: f.relativePath, name: f.name, extension: f.extension }))));
-    fs.writeFileSync(path.join(aiDir, "repo-map.json"), JSON.stringify(repoMapData, null, 2));
-    console.log("   ✅ repo-map.json");
+    fs.writeFileSync(path.join(aiDir, "repo_map.json"), JSON.stringify(repoMapData, null, 2));
+    console.log("   ✅ repo_map.json");
     
     // module-graph.json
     const { generateModuleGraph } = await import("../core/moduleGraph.js");
@@ -1183,7 +1205,7 @@ Examples:
     args.shift();
     
     let rootDir = process.cwd();
-    let aiDir = path.join(rootDir, "ai");
+    let aiDir = path.join(rootDir, "ai-context");
     let limit = 50;
     let showActivity = false;
     let showJson = false;
@@ -1194,7 +1216,7 @@ Examples:
         case "--root":
         case "-r":
           rootDir = args[++i];
-          aiDir = path.join(rootDir, "ai");
+          aiDir = path.join(rootDir, "ai-context");
           break;
         case "--limit":
         case "-n":
@@ -1290,7 +1312,7 @@ Examples:
     args.shift();
     
     let rootDir = process.cwd();
-    let aiDir = path.join(rootDir, "ai");
+    let aiDir = path.join(rootDir, "ai-context");
     let showJson = false;
     let showStats = false;
     
@@ -1300,7 +1322,7 @@ Examples:
         case "--root":
         case "-r":
           rootDir = args[++i];
-          aiDir = path.join(rootDir, "ai");
+          aiDir = path.join(rootDir, "ai-context");
           break;
         case "--json":
           showJson = true;
@@ -1373,7 +1395,7 @@ Examples:
     args.shift();
     
     let rootDir = process.cwd();
-    let aiDir = path.join(rootDir, "ai");
+    let aiDir = path.join(rootDir, "ai-context");
     let useGit = true;
     let showJson = false;
     
@@ -1383,7 +1405,7 @@ Examples:
         case "--root":
         case "-r":
           rootDir = args[++i];
-          aiDir = path.join(rootDir, "ai");
+          aiDir = path.join(rootDir, "ai-context");
           break;
         case "--no-git":
           useGit = false;

package/src/commands/doctor.ts

@@ -30,7 +30,7 @@ export async function runDoctor(rootDir: string, fixMode: boolean): Promise<Doct
   const largeFiles = scanResult.files.filter(f => { try { return fs.statSync(f.path).size > 1024 * 1024; } catch { return false; } });
   checks.push({ name: 'Large files', status: largeFiles.length === 0 ? 'pass' : 'warn', message: largeFiles.length === 0 ? 'No large files' : `${largeFiles.length} large files` });
 
-  const aiDir = path.join(rootDir, "ai");
+  const aiDir = path.join(rootDir, "ai-context");
   checks.push({ name: 'AI directory', status: fs.existsSync(aiDir) ? 'pass' : 'warn', message: fs.existsSync(aiDir) ? 'Found' : 'Not found' });
   checks.push({ name: 'Semantic index', status: fs.existsSync(path.join(aiDir, "index")) ? 'pass' : 'warn', message: fs.existsSync(path.join(aiDir, "index")) ? 'Found' : 'Not found' });
   checks.push({ name: 'Module graph', status: fs.existsSync(path.join(aiDir, "graph", "module-graph.json")) ? 'pass' : 'warn', message: 'Run ai-first explore all' });

package/src/commands/explore.ts

@@ -6,7 +6,7 @@ import { generateModuleGraph } from "../core/moduleGraph.js";
 export interface ExploreResult { success: boolean; modules?: string[]; dependencies?: string[]; error?: string; }
 
 export async function runExplore(rootDir: string, moduleName: string | undefined): Promise<ExploreResult> {
-  const aiDir = path.join(rootDir, "ai");
+  const aiDir = path.join(rootDir, "ai-context");
   const graphFile = path.join(aiDir, "graph", "module-graph.json");
 
   if (!moduleName || moduleName === 'all') {

package/src/core/ccp.ts

@@ -147,7 +147,7 @@ export function createCCP(
     include?: string[];
   } = {}
 ): { success: boolean; path: string; error?: string } {
-  const aiDir = path.join(rootDir, "ai");
+  const aiDir = path.join(rootDir, "ai-context");
   const ccpDir = path.join(aiDir, "ccp", name);
   
   try {

package/src/core/embeddings.ts

@@ -1,6 +1,5 @@
-import fs from "fs";
-import path from "path";
 import crypto from "crypto";
+import { Database } from "sql.js";
 import { Chunk } from "./chunker.js";
 
 export interface Embedding {
@@ -24,6 +23,26 @@ export interface EmbeddingsIndex {
   embeddings: Embedding[];
 }
 
+/**
+ * Create embeddings table in the database
+ */
+export function createEmbeddingsTable(db: Database): void {
+  db.run(`
+    CREATE TABLE IF NOT EXISTS embeddings (
+      id TEXT PRIMARY KEY,
+      vector TEXT NOT NULL,
+      metadata TEXT NOT NULL,
+      chunk_id TEXT,
+      file_path TEXT,
+      start_line INTEGER,
+      end_line INTEGER,
+      type TEXT,
+      language TEXT
+    )
+  `);
+  db.run("CREATE INDEX IF NOT EXISTS idx_embeddings_file_path ON embeddings(file_path)");
+}
+
 /**
  * Generate a simple hash-based embedding (for testing/fallback)
  * This creates a deterministic embedding based on content
@@ -86,40 +105,72 @@ export async function generateEmbeddings(
 }
 
 /**
- * Save embeddings to file
+ * Save embeddings to SQLite database
  */
 export function saveEmbeddings(
+  db: Database,
   embeddings: Embedding[],
-  aiDir: string,
   model: string,
   dimensions: number
 ): void {
-  const index: EmbeddingsIndex = {
-    version: "1.0.0",
-    model,
-    dimensions,
-    generated: new Date().toISOString(),
-    embeddings
-  };
+  db.run("DELETE FROM embeddings");
   
-  const embeddingsPath = path.join(aiDir, "embeddings.json");
-  fs.writeFileSync(embeddingsPath, JSON.stringify(index, null, 2));
-  console.log(`   ✅ Saved ${embeddings.length} embeddings to ${embeddingsPath}`);
+  for (const embedding of embeddings) {
+    const vectorJson = JSON.stringify(embedding.vector);
+    const metadataJson = JSON.stringify(embedding.metadata);
+    
+    db.run(
+      `INSERT INTO embeddings (id, vector, metadata, chunk_id, file_path, start_line, end_line, type, language) 
+       VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,
+      [
+        embedding.id,
+        vectorJson,
+        metadataJson,
+        embedding.metadata.chunkId,
+        embedding.metadata.filePath,
+        embedding.metadata.startLine,
+        embedding.metadata.endLine,
+        embedding.metadata.type,
+        embedding.metadata.language
+      ]
+    );
+  }
+  
+  console.log(`   ✅ Saved ${embeddings.length} embeddings to database`);
 }
 
 /**
- * Load embeddings from file
+ * Load embeddings from SQLite database
  */
-export function loadEmbeddings(aiDir: string): EmbeddingsIndex | null {
-  const embeddingsPath = path.join(aiDir, "embeddings.json");
-  
-  if (!fs.existsSync(embeddingsPath)) {
-    return null;
-  }
-  
+export function loadEmbeddings(db: Database): EmbeddingsIndex | null {
   try {
-    const data = fs.readFileSync(embeddingsPath, "utf-8");
-    return JSON.parse(data);
+    const result = db.exec("SELECT id, vector, metadata FROM embeddings");
+    
+    if (result.length === 0 || result[0].values.length === 0) {
+      return null;
+    }
+    
+    const embeddings: Embedding[] = [];
+    
+    for (const row of result[0].values) {
+      const id = row[0] as string;
+      const vectorJson = row[1] as string;
+      const metadataJson = row[2] as string;
+      
+      embeddings.push({
+        id,
+        vector: JSON.parse(vectorJson),
+        metadata: JSON.parse(metadataJson)
+      });
+    }
+    
+    return {
+      version: "1.0.0",
+      model: "simple-fallback",
+      dimensions: embeddings.length > 0 ? embeddings[0].vector.length : 384,
+      generated: new Date().toISOString(),
+      embeddings
+    };
   } catch {
     return null;
   }
@@ -143,30 +194,45 @@ function cosineSimilarity(a: number[], b: number[]): number {
 }
 
 /**
- * Search embeddings for similar chunks
+ * Search embeddings for similar chunks using SQLite database
  */
 export function searchEmbeddings(
+  db: Database,
   query: string,
-  embeddingsIndex: EmbeddingsIndex,
+  dimensions: number = 384,
   topK: number = 5
 ): { chunkId: string; filePath: string; score: number; content: string }[] {
-  // Generate embedding for query
-  const queryEmbedding = generateSimpleEmbedding(query, embeddingsIndex.dimensions);
+  const queryEmbedding = generateSimpleEmbedding(query, dimensions);
   
-  // Calculate similarities
-  const results = embeddingsIndex.embeddings.map(embedding => ({
-    chunkId: embedding.id,
-    filePath: embedding.metadata.filePath,
-    score: cosineSimilarity(queryEmbedding, embedding.vector),
-    startLine: embedding.metadata.startLine,
-    endLine: embedding.metadata.endLine,
-    type: embedding.metadata.type
-  }));
+  const result = db.exec("SELECT id, vector, metadata FROM embeddings");
+  
+  if (result.length === 0 || result[0].values.length === 0) {
+    return [];
+  }
+  
+  const searchResults: { chunkId: string; filePath: string; score: number; startLine: number; endLine: number; type: string }[] = [];
+  
+  for (const row of result[0].values) {
+    const id = row[0] as string;
+    const vectorJson = row[1] as string;
+    const metadataJson = row[2] as string;
+    
+    const vector = JSON.parse(vectorJson);
+    const metadata = JSON.parse(metadataJson);
+    
+    searchResults.push({
+      chunkId: id,
+      filePath: metadata.filePath,
+      score: cosineSimilarity(queryEmbedding, vector),
+      startLine: metadata.startLine,
+      endLine: metadata.endLine,
+      type: metadata.type
+    });
+  }
   
-  // Sort by score and return top K
-  results.sort((a, b) => b.score - a.score);
+  searchResults.sort((a, b) => b.score - a.score);
   
-  return results.slice(0, topK).map(r => ({
+  return searchResults.slice(0, topK).map(r => ({
     chunkId: r.chunkId,
     filePath: r.filePath,
     score: r.score,

package/src/core/gitAnalyzer.ts

@@ -240,7 +240,7 @@ function loadFlows(aiDir: string): Map<string, string[]> {
  * Map changed files to features
  */
 export function mapFilesToFeatures(rootDir: string, files: string[]): string[] {
-  const aiDir = path.join(rootDir, "ai");
+  const aiDir = path.join(rootDir, "ai-context");
   const featuresMap = loadFeatures(aiDir);
   
   const featureSet = new Set<string>();
@@ -267,7 +267,7 @@ export function mapFilesToFeatures(rootDir: string, files: string[]): string[] {
  * Map changed files to flows
  */
 export function mapFilesToFlows(rootDir: string, files: string[]): string[] {
-  const aiDir = path.join(rootDir, "ai");
+  const aiDir = path.join(rootDir, "ai-context");
   const flowsMap = loadFlows(aiDir);
   
   const flowSet = new Set<string>();
@@ -354,7 +354,7 @@ export function generateGitContext(rootDir: string, aiDir?: string): {
   recentFlows: string[];
   activity: GitActivity | null;
 } {
-  const targetAiDir = aiDir || path.join(rootDir, "ai");
+  const targetAiDir = aiDir || path.join(rootDir, "ai-context");
   const gitDir = path.join(targetAiDir, "git");
   
   ensureDir(gitDir);