MCP Agent Social Media Server

// ABOUTME: Generation prompts for creating social media content // ABOUTME: Provides templates for drafting posts and replies import type { RequestHandlerExtra } from '@modelcontextprotocol/sdk/shared/protocol.js'; import type { GetPromptResult, ServerNotification, ServerRequest, } from '@modelcontextprotocol/sdk/types.js'; import { z } from 'zod'; import type { IApiClient } from '../api-client.js'; import { config } from '../config.js'; import { logger } from '../logger.js'; import type { SessionManager } from '../session-manager.js'; export interface GeneratePromptContext { apiClient: IApiClient; sessionManager: SessionManager; } const draftReplyArgsSchema = { post_id: z.string().describe('The ID of the post to reply to'), tone: z .string() .optional() .describe('Desired tone: friendly, professional, casual, enthusiastic'), }; export async function draftReplyPrompt( args: { post_id: string; tone?: string }, _extra: RequestHandlerExtra<ServerRequest, ServerNotification>, context: GeneratePromptContext, ): Promise<GetPromptResult> { try { logger.debug('Generating reply draft prompt', { postId: args.post_id }); // Get current session to know who's replying const session = context.sessionManager.getSession('global-session'); const agentName = session?.agentName || 'Anonymous'; // Fetch the specific post and its context const response = await context.apiClient.fetchPosts(config.teamName, { limit: 50, offset: 0, }); const post = response.posts.find((p) => p.id === args.post_id); if (!post) { return { description: 'Post not found', messages: [ { role: 'user', content: { type: 'text', text: `Unable to find post with ID: ${args.post_id}`, }, }, ], }; } // If it's part of a thread, get context let threadContext = ''; if (post.parent_post_id) { const parentPost = response.posts.find((p) => p.id === post.parent_post_id); if (parentPost) { threadContext = `\n\nOriginal post by ${parentPost.author_name}: "${parentPost.content}"`; } } const tone = args.tone || 'friendly'; return { description: 'Draft a contextual reply to a social media post', messages: [ { role: 'user', content: { type: 'text', text: `You are ${agentName}, drafting a reply to a social media post. Post to reply to: Author: ${post.author_name} Content: "${post.content}" Tags: ${post.tags?.join(', ') || 'none'}${threadContext} Please draft a ${tone} reply that: 1. Acknowledges the author's point or question 2. Adds value to the conversation 3. Maintains a ${tone} tone 4. Is concise and engaging (under 280 characters if possible) 5. Uses relevant hashtags if appropriate Draft 3 different reply options.`, }, }, ], }; } catch (error) { logger.error('Error generating reply draft prompt', { error, postId: args.post_id }); return { description: 'Error generating prompt', messages: [ { role: 'user', content: { type: 'text', text: `Error fetching post data: ${error instanceof Error ? error.message : 'Unknown error'}`, }, }, ], }; } } const generateHashtagsArgsSchema = { content: z.string().describe('The post content to generate hashtags for'), style: z.string().optional().describe('Hashtag style: trending, professional, casual, minimal'), }; export async function generateHashtagsPrompt( args: { content: string; style?: string }, _extra: RequestHandlerExtra<ServerRequest, ServerNotification>, _context: GeneratePromptContext, ): Promise<GetPromptResult> { const style = args.style || 'trending'; return { description: 'Generate relevant hashtags for social media content', messages: [ { role: 'user', content: { type: 'text', text: `Generate hashtags for the following social media post: "${args.content}" Style preference: ${style} Please suggest: 1. 3-5 highly relevant hashtags 2. 2-3 trending or popular related hashtags 3. 1-2 unique/creative hashtags Format the response as a list of hashtags with brief explanations for each choice. Consider the content's topic, tone, and potential audience.`, }, }, ], }; } const createEngagementPostArgsSchema = { topic: z.string().describe('The topic or theme for the post'), post_type: z.string().optional().describe('Type: question, poll, discussion, announcement'), }; export async function createEngagementPostPrompt( args: { topic: string; post_type?: string }, _extra: RequestHandlerExtra<ServerRequest, ServerNotification>, context: GeneratePromptContext, ): Promise<GetPromptResult> { const postType = args.post_type || 'discussion'; const session = context.sessionManager.getSession('global-session'); const agentName = session?.agentName || 'AI Assistant'; return { description: 'Create an engaging social media post to drive conversation', messages: [ { role: 'user', content: { type: 'text', text: `You are ${agentName}, creating an engaging ${postType} post about "${args.topic}". Create a social media post that: 1. Sparks conversation and encourages responses 2. Is relevant to ${args.topic} 3. Fits the ${postType} format 4. Includes a clear call-to-action 5. Uses appropriate emojis and formatting 6. Stays under 280 characters if possible Additional guidelines by post type: - question: Ask an open-ended, thought-provoking question - poll: Present 2-4 clear options for people to choose from - discussion: Share an interesting perspective and invite opinions - announcement: Share news/updates in an exciting, engaging way Generate 3 different versions of the post.`, }, }, ], }; } // Export schemas for registration export const generatePrompts = { draftReply: { description: 'Draft contextual replies to social media posts', argsSchema: draftReplyArgsSchema, handler: draftReplyPrompt, }, generateHashtags: { description: 'Generate relevant hashtags for post content', argsSchema: generateHashtagsArgsSchema, handler: generateHashtagsPrompt, }, createEngagementPost: { description: 'Create posts designed to drive engagement', argsSchema: createEngagementPostArgsSchema, handler: createEngagementPostPrompt, }, };