Codex MCP Server

import { z } from 'zod'; import { UnifiedTool } from './registry.js'; import { Logger } from '../utils/logger.js'; import { executeCodexCLI } from '../utils/codexExecutor.js'; function buildBrainstormPrompt(config: { prompt: string; methodology: string; domain?: string; constraints?: string; existingContext?: string; ideaCount: number; includeAnalysis: boolean; }): string { const { prompt, methodology, domain, constraints, existingContext, ideaCount, includeAnalysis } = config; // Select methodology framework let frameworkInstructions = getMethodologyInstructions(methodology, domain); let enhancedPrompt = `# BRAINSTORMING SESSION ## Challenge: ${prompt} ## Framework ${frameworkInstructions} ## Context ${domain ? `Domain: ${domain}` : ''} ${constraints ? `Constraints: ${constraints}` : ''} ${existingContext ? `Background: ${existingContext}` : ''} ## Requirements Generate ${ideaCount} actionable ideas. Keep descriptions concise (2-3 sentences max). ${ includeAnalysis ? `## Analysis Rate each: Feasibility (1-5), Impact (1-5), Innovation (1-5)` : '' } ## Format ### Idea [N]: [Name] Description: [2-3 sentences] ${includeAnalysis ? 'Ratings: F:[1-5] I:[1-5] N:[1-5]' : ''} Begin:`; return enhancedPrompt; } /** * Returns methodology-specific instructions for structured brainstorming */ function getMethodologyInstructions(methodology: string, domain?: string): string { const methodologies: Record<string, string> = { divergent: `**Divergent Thinking Approach:** - Generate maximum quantity of ideas without self-censoring - Build on wild or seemingly impractical ideas - Combine unrelated concepts for unexpected solutions - Use "Yes, and..." thinking to expand each concept - Postpone evaluation until all ideas are generated`, convergent: `**Convergent Thinking Approach:** - Focus on refining and improving existing concepts - Synthesize related ideas into stronger solutions - Apply critical evaluation criteria - Prioritize based on feasibility and impact - Develop implementation pathways for top ideas`, scamper: `**SCAMPER Creative Triggers:** - **Substitute:** What can be substituted or replaced? - **Combine:** What can be combined or merged? - **Adapt:** What can be adapted from other domains? - **Modify:** What can be magnified, minimized, or altered? - **Put to other use:** How else can this be used? - **Eliminate:** What can be removed or simplified? - **Reverse:** What can be rearranged or reversed?`, 'design-thinking': `**Human-Centered Design Thinking:** - **Empathize:** Consider user needs, pain points, and contexts - **Define:** Frame problems from user perspective - **Ideate:** Generate user-focused solutions - **Consider Journey:** Think through complete user experience - **Prototype Mindset:** Focus on testable, iterative concepts`, lateral: `**Lateral Thinking Approach:** - Make unexpected connections between unrelated fields - Challenge fundamental assumptions - Use random word association to trigger new directions - Apply metaphors and analogies from other domains - Reverse conventional thinking patterns`, auto: `**AI-Optimized Approach:** ${domain ? `Given the ${domain} domain, I'll apply the most effective combination of:` : "I'll intelligently combine multiple methodologies:"} - Divergent exploration with domain-specific knowledge - SCAMPER triggers and lateral thinking - Human-centered perspective for practical value`, }; return methodologies[methodology] || methodologies['auto']; } const brainstormArgsSchema = z.object({ prompt: z.string().min(1).describe('Brainstorming challenge or question'), model: z .string() .optional() .describe( 'Model: gpt-5-codex (default), gpt-5, o3, o4-mini, codex-1, codex-mini-latest, gpt-4.1' ), approvalPolicy: z .enum(['never', 'on-request', 'on-failure', 'untrusted']) .optional() .describe('Approval: never, on-request, on-failure, untrusted'), sandboxMode: z .enum(['read-only', 'workspace-write', 'danger-full-access']) .optional() .describe('Access: read-only, workspace-write, danger-full-access'), fullAuto: z.boolean().optional().describe('Full automation mode'), yolo: z.boolean().optional().describe('⚠️ Bypass all safety (dangerous)'), cd: z.string().optional().describe('Working directory'), methodology: z .enum(['divergent', 'convergent', 'scamper', 'design-thinking', 'lateral', 'auto']) .default('auto') .describe( 'Framework: divergent, convergent, scamper, design-thinking, lateral, auto (default)' ), domain: z .string() .optional() .describe('Domain: software, business, creative, research, product, marketing, etc.'), constraints: z.string().optional().describe('Limitations: budget, time, technical, legal, etc.'), existingContext: z.string().optional().describe('Background info or previous attempts'), ideaCount: z .number() .int() .positive() .default(12) .describe('Number of ideas (default: 12, range: 5-30)'), includeAnalysis: z.boolean().default(true).describe('Include feasibility/impact analysis'), search: z .boolean() .optional() .describe('Enable web search for research (activates web_search_request feature)'), oss: z.boolean().optional().describe('Use local Ollama server'), enableFeatures: z.array(z.string()).optional().describe('Enable feature flags'), disableFeatures: z.array(z.string()).optional().describe('Disable feature flags'), }); export const brainstormTool: UnifiedTool = { name: 'brainstorm', description: 'Generate creative ideas using structured frameworks with domain context and feasibility analysis.', zodSchema: brainstormArgsSchema, prompt: { description: 'Create structured brainstorming with chosen methodology and analysis', }, category: 'utility', execute: async (args, onProgress) => { const { prompt, model, approvalPolicy, sandboxMode, fullAuto, yolo, cd, methodology = 'auto', domain, constraints, existingContext, ideaCount = 12, includeAnalysis = true, search, oss, enableFeatures, disableFeatures, } = args; if (!prompt?.trim()) { throw new Error('You must provide a valid brainstorming challenge or question to explore'); } let enhancedPrompt = buildBrainstormPrompt({ prompt: prompt.trim() as string, methodology: methodology as string, domain: domain as string | undefined, constraints: constraints as string | undefined, existingContext: existingContext as string | undefined, ideaCount: ideaCount as number, includeAnalysis: includeAnalysis as boolean, }); Logger.debug( `Brainstorm: Using methodology '${methodology}' for domain '${domain || 'general'}'` ); // Report progress to user onProgress?.(`Generating ${ideaCount} ideas via ${methodology} methodology...`); // Execute with Codex (non-interactive) return await executeCodexCLI( enhancedPrompt, { model: model as string | undefined, fullAuto: Boolean(fullAuto), approvalPolicy: approvalPolicy as any, sandboxMode: sandboxMode as any, yolo: Boolean(yolo), cd: cd as string | undefined, search: search as boolean, oss: oss as boolean, enableFeatures: enableFeatures as string[], disableFeatures: disableFeatures as string[], }, onProgress ); }, };