claudex-setup 1.3.0 → 1.4.0

package/bin/cli.js

@@ -49,6 +49,12 @@ async function main() {
     dir: process.cwd()
   };
 
+  if (!require('fs').existsSync(options.dir)) {
+    console.error(`\n  Error: Directory not found: ${options.dir}`);
+    console.error('  Run claudex-setup from inside your project directory.\n');
+    process.exit(1);
+  }
+
   try {
     if (command === 'badge') {
       const { getBadgeMarkdown } = require('../src/badge');

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claudex-setup",
-  "version": "1.3.0",
+  "version": "1.4.0",
   "description": "Audit and optimize any project for Claude Code. Powered by 1107 verified techniques.",
   "main": "src/index.js",
   "bin": {

package/src/audit.js

@@ -97,7 +97,18 @@ async function audit(options) {
   }
 
   if (options.json) {
-    console.log(JSON.stringify({ score, stacks, passed: passed.length, failed: failed.length, results }, null, 2));
+    const { version } = require('../package.json');
+    console.log(JSON.stringify({
+      version,
+      timestamp: new Date().toISOString(),
+      score,
+      stacks,
+      passed: passed.length,
+      failed: failed.length,
+      skipped: skipped.length,
+      checkCount: applicable.length,
+      results
+    }, null, 2));
     return { score, passed: passed.length, failed: failed.length, stacks, results };
   }
 

package/src/setup.js

@@ -29,8 +29,7 @@ function detectScripts(ctx) {
 // Helper: detect key dependencies and generate guidelines
 // ============================================================
 function detectDependencies(ctx) {
-  const pkg = ctx.jsonFile('package.json');
-  if (!pkg) return [];
+  const pkg = ctx.jsonFile('package.json') || {};
   const allDeps = { ...(pkg.dependencies || {}), ...(pkg.devDependencies || {}) };
   const guidelines = [];
 
@@ -153,6 +152,50 @@ function detectDependencies(ctx) {
   if (reqTxt.includes('redis')) {
     guidelines.push('- Redis is available for caching and task queues');
   }
+  if (reqTxt.includes('langchain')) {
+    guidelines.push('- Use LangChain for chain/agent orchestration. Define chains in chains/ directory');
+  }
+  if (reqTxt.includes('openai')) {
+    guidelines.push('- OpenAI SDK available. Use structured outputs where possible');
+  }
+  if (reqTxt.includes('anthropic')) {
+    guidelines.push('- Anthropic SDK available. Prefer Claude for complex reasoning tasks');
+  }
+  if (reqTxt.includes('chromadb')) {
+    guidelines.push('- Use ChromaDB for local vector storage. Persist collections to disk');
+  }
+  if (reqTxt.includes('pinecone')) {
+    guidelines.push('- Use Pinecone for production vector search. Define index schemas upfront');
+  }
+  if (reqTxt.includes('mlflow')) {
+    guidelines.push('- Use MLflow for experiment tracking. Log all model parameters and metrics');
+  }
+  if (reqTxt.includes('wandb')) {
+    guidelines.push('- Use Weights & Biases for experiment tracking and visualization');
+  }
+  if (reqTxt.includes('transformers')) {
+    guidelines.push('- HuggingFace Transformers available. Use AutoModel/AutoTokenizer for loading');
+  }
+
+  // JS AI/ML/Cloud deps
+  if (allDeps['@anthropic-ai/sdk']) {
+    guidelines.push('- Anthropic SDK configured. Use Messages API with structured tool_use for agents');
+  }
+  if (allDeps['openai']) {
+    guidelines.push('- OpenAI SDK available. Use structured outputs and function calling');
+  }
+  if (allDeps['@modelcontextprotocol/sdk']) {
+    guidelines.push('- MCP SDK available. Build MCP servers with stdio transport');
+  }
+  if (allDeps['langchain'] || allDeps['@langchain/core']) {
+    guidelines.push('- LangChain available. Use LCEL for chain composition');
+  }
+  if (allDeps['@aws-sdk/client-s3'] || allDeps['@aws-sdk/client-dynamodb']) {
+    guidelines.push('- AWS SDK v3 configured. Use modular imports, not aws-sdk v2');
+  }
+  if (allDeps['@aws-cdk/aws-lambda'] || allDeps['aws-cdk-lib']) {
+    guidelines.push('- AWS CDK available. Define stacks in lib/, constructs as separate classes');
+  }
 
   return guidelines;
 }
@@ -564,6 +607,12 @@ Before completing any task, confirm:
 ${verificationSteps.join('\n')}
 </verification>
 
+## Context Management
+- Use /compact when context gets large (above 50% capacity)
+- Prefer focused sessions — one task per conversation
+- If a session gets too long, start fresh with /clear
+- Use subagents for research tasks to keep main context clean
+
 ## Workflow
 - Verify changes with tests before committing
 - Use descriptive commit messages (why, not what)