codex-cli-architect-mcp/dist/tools/consult.tool.js

MCP server for OpenAI Codex CLI - Enable AI-powered coding assistance through Model Context Protocol

import { z } from 'zod';
import { DEFAULT_MODEL } from '../constants.js';
import { executeCodexCommand } from '../utils/codexExecutor.js';
export const consultSchema = z.object({
    topic: z
        .enum(['architecture', 'design-pattern', 'tech-stack', 'implementation'])
        .describe('Type of consultation'),
    question: z.string().min(1).describe('Your technical question or design problem'),
    context: z.string().optional().describe('Additional context about your situation'),
    constraints: z.array(z.string()).optional().describe('Technical constraints or requirements'),
    model: z.string().optional().describe('AI model to use'),
});
export const codexConsultTool = {
    name: 'codex_consult',
    description: 'Technical consultation for architecture decisions, design patterns, technology choices, and implementation strategies',
    schema: consultSchema,
    async execute(args) {
        const { topic, question, context, constraints = [], model = DEFAULT_MODEL } = args;
        let prompt = `As a senior technical consultant, provide expert advice on the following ${topic} question:\n\n`;
        prompt += `Question: ${question}\n`;
        if (context) {
            prompt += `Context: ${context}\n`;
        }
        if (constraints.length > 0) {
            prompt += `Constraints: ${constraints.join(', ')}\n`;
        }
        const result = await executeCodexCommand(prompt, { model });
        if (!result.success) {
            throw new Error(result.error || 'Failed to provide consultation');
        }
        return result.output || 'No consultation response generated';
    },
};
//# sourceMappingURL=consult.tool.js.map