Codex MCP Server

# Advanced Usage This guide covers advanced patterns and techniques for using Codex MCP Tool effectively in complex scenarios. ## Working with Large Codebases ### Chunked Analysis Strategy When analyzing large codebases, break down the analysis into manageable chunks: ```javascript // Step 1: Get overview { "name": "ask-codex", "arguments": { "prompt": "provide high-level architecture overview of @src/", "model": "o4-mini" } } // Step 2: Deep dive into specific modules { "name": "ask-codex", "arguments": { "prompt": "analyze @src/core/ in detail", "model": "gpt-5" } } // Step 3: Cross-module dependencies { "name": "ask-codex", "arguments": { "prompt": "map dependencies between @src/core/ and @src/utils/", "model": "o3" } } ``` ### Memory-Efficient Processing For very large files or directories: ```javascript // Use changeMode with chunking { "name": "ask-codex", "arguments": { "prompt": "refactor all TypeScript files in @src/", "changeMode": true, "model": "gpt-5" } } // Fetch chunks as needed { "name": "fetch-chunk", "arguments": { "cacheKey": "refactor-xyz", "chunkIndex": 1 } } ``` ## Complex Refactoring Patterns ### Multi-Stage Refactoring Execute refactoring in controlled stages: ```javascript // Stage 1: Analysis const analysis = { name: 'ask-codex', arguments: { prompt: 'identify all code smells and anti-patterns in @src/', sandboxMode: 'read-only', }, }; // Stage 2: Planning const plan = { name: 'brainstorm', arguments: { prompt: 'create refactoring plan for identified issues', methodology: 'convergent', existingContext: '[analysis results]', }, }; // Stage 3: Implementation const implement = { name: 'ask-codex', arguments: { prompt: 'implement refactoring plan step by step', changeMode: true, sandboxMode: 'workspace-write', approvalPolicy: 'on-request', }, }; ``` ### Incremental Migration For framework or library migrations: ```javascript // Migrate component by component const components = ['Button', 'Form', 'Modal', 'Table']; for (const component of components) { // Analyze current implementation await mcp.call('ask-codex', { prompt: `analyze @src/components/${component}.tsx`, sandboxMode: 'read-only', }); // Generate migration await mcp.call('ask-codex', { prompt: `migrate ${component} from v4 to v5 API`, changeMode: true, approvalPolicy: 'on-request', }); // Verify migration await mcp.call('ask-codex', { prompt: `verify ${component} migration is complete`, sandboxMode: 'read-only', }); } ``` ## Advanced Brainstorming Techniques ### Iterative Refinement Build on previous brainstorming sessions: ```javascript // Round 1: Generate ideas const round1 = { name: 'brainstorm', arguments: { prompt: 'innovative features for developer tool', methodology: 'divergent', ideaCount: 30, }, }; // Round 2: Filter and refine const round2 = { name: 'brainstorm', arguments: { prompt: 'refine top 10 ideas from previous session', methodology: 'convergent', existingContext: '[round1 results]', constraints: 'must be implementable in 2 weeks', }, }; // Round 3: Implementation planning const round3 = { name: 'brainstorm', arguments: { prompt: 'create implementation plan for top 3 ideas', methodology: 'design-thinking', existingContext: '[round2 results]', includeAnalysis: true, }, }; ``` ### Cross-Domain Innovation Combine multiple domains for creative solutions: ```javascript { "name": "brainstorm", "arguments": { "prompt": "apply gaming mechanics to productivity app", "methodology": "lateral", "domain": "gaming,productivity,psychology", "ideaCount": 20 } } ``` ## Automation Workflows ### CI/CD Integration Integrate Codex MCP Tool into your CI/CD pipeline: ```yaml # .github/workflows/code-review.yml name: Automated Code Review on: pull_request: types: [opened, synchronize] jobs: review: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - name: Setup Codex MCP run: npm install -g @trishchuk/codex-mcp-tool - name: Security Review run: | echo '{ "method": "tools/call", "params": { "name": "ask-codex", "arguments": { "prompt": "security audit @src/", "model": "gpt-5", "sandboxMode": "read-only" } } }' | npx @trishchuk/codex-mcp-tool - name: Performance Review run: | echo '{ "method": "tools/call", "params": { "name": "ask-codex", "arguments": { "prompt": "identify performance bottlenecks @src/", "model": "o3" } } }' | npx @trishchuk/codex-mcp-tool ``` ### Scheduled Maintenance Automate regular code maintenance tasks: ```javascript // Weekly technical debt review const weeklyReview = async () => { // Identify technical debt const debt = await mcp.call('ask-codex', { prompt: 'identify technical debt in @src/', model: 'gpt-5', }); // Generate improvement plan const plan = await mcp.call('brainstorm', { prompt: 'prioritize technical debt fixes', existingContext: debt, constraints: '4 hours per week available', }); // Create tickets await createGitHubIssues(plan); }; ``` ## Performance Optimization ### Model Selection Strategy Choose the right model for the task: ```javascript const selectModel = task => { switch (task.type) { case 'quick-analysis': return 'o4-mini'; // Fast, cost-effective case 'complex-reasoning': return 'o3'; // Advanced reasoning case 'large-refactoring': return 'gpt-5'; // Maximum context default: return 'o4-mini'; } }; // Use in practice const model = selectModel({ type: 'complex-reasoning' }); await mcp.call('ask-codex', { prompt: 'your task', model: model, }); ``` ### Parallel Processing Execute independent tasks in parallel: ```javascript // Parallel analysis of different modules const modules = ['auth', 'api', 'database', 'ui']; const analyses = await Promise.all( modules.map(module => mcp.call('ask-codex', { prompt: `analyze @src/${module}/`, model: 'o4-mini', }) ) ); // Combine results const summary = await mcp.call('ask-codex', { prompt: 'synthesize module analyses into system overview', existingContext: analyses.join('\n'), model: 'gpt-5', }); ``` ## Custom Tool Chains ### Building Complex Workflows Chain multiple tools for sophisticated operations: ```javascript class CodexWorkflow { async reviewAndRefactor(path) { // Step 1: Initial review const review = await this.review(path); // Step 2: Generate improvement ideas const ideas = await this.brainstorm(review); // Step 3: Implement top improvements const changes = await this.implement(ideas); // Step 4: Verify changes const verification = await this.verify(changes); return { review, ideas, changes, verification }; } async review(path) { return mcp.call('ask-codex', { prompt: `comprehensive review of @${path}`, model: 'gpt-5', sandboxMode: 'read-only', }); } async brainstorm(context) { return mcp.call('brainstorm', { prompt: 'improvement suggestions', existingContext: context, methodology: 'scamper', ideaCount: 10, }); } async implement(ideas) { return mcp.call('ask-codex', { prompt: `implement top 3 improvements: ${ideas}`, changeMode: true, sandboxMode: 'workspace-write', approvalPolicy: 'on-request', }); } async verify(changes) { return mcp.call('ask-codex', { prompt: 'verify changes maintain functionality', sandboxMode: 'read-only', }); } } ``` ## Security Best Practices ### Secure Code Review Implement security-focused reviews: ```javascript { "name": "ask-codex", "arguments": { "prompt": `Check @src/ for: - SQL injection vulnerabilities - XSS vulnerabilities - Authentication bypasses - Insecure deserialization - Sensitive data exposure - Security misconfigurations - Using components with known vulnerabilities - Insufficient logging and monitoring`, "model": "gpt-5", "sandboxMode": "read-only" } } ``` ### Compliance Checking Ensure code meets compliance requirements: ```javascript { "name": "ask-codex", "arguments": { "prompt": "verify @src/ compliance with GDPR, HIPAA, and SOC 2", "model": "gpt-5", "sandboxMode": "read-only" } } ``` ## Debugging Complex Issues ### Root Cause Analysis Systematic debugging approach: ```javascript // Step 1: Reproduce issue const reproduce = { name: 'ask-codex', arguments: { prompt: 'trace execution path for [error scenario] in @src/', model: 'o3', }, }; // Step 2: Identify root cause const rootCause = { name: 'ask-codex', arguments: { prompt: 'identify root cause of [error] based on trace', model: 'gpt-5', }, }; // Step 3: Generate fixes const fixes = { name: 'brainstorm', arguments: { prompt: 'possible fixes for [root cause]', methodology: 'convergent', includeAnalysis: true, }, }; ``` ## Related Resources - [Basic Usage](./basic-usage.md) - [API Reference](../api/tools/ask-codex.md) - [Troubleshooting](../resources/troubleshooting.md) - [FAQ](../resources/faq.md)