How to Build an AI Social Media Manager with Claude Code
Disclosure: some links in this post are affiliate links. If you buy through them I may earn a commission at no extra cost to you. I only recommend tools I actually use. Full disclosure.
This guide walks you through how to build a fully functional Claude Code Social Media system that posts, schedules, and repurposes content across every platform you use.How Claude Code Social Media Management Works
Most AI social media tools generate captions in a vacuum. They have no idea how you sound, what platforms you post on, or what your content calendar looks like. Claude Code takes a different approach. You give it skill files (markdown SOPs that define your voice, platform rules, and posting preferences), connect your social accounts through an MCP integration, and then manage everything from a single terminal interface. No switching between tabs. No copy-pasting into five different scheduling tools. The setup involves three core pieces: Claude Code running inside VS Code (or terminal or desktop app), a social media MCP connection through a tool called Blotato, and skill files that train the AI on your exact brand voice. Once those are connected, you can repurpose a YouTube video into platform-specific posts, publish directly to X, or schedule content weeks in advance on LinkedIn, Instagram, TikTok, and more.
Connecting Your Social Media Accounts to Claude Code
You have probably tried connecting APIs manually before. It is tedious and breaks constantly. Blotato simplifies this by acting as the bridge between Claude Code and your social platforms. Plans start at $29/month with a 30-day money-back guarantee. Once you are on at least the Starter plan, you connect each social account (X, LinkedIn, Facebook, TikTok, Instagram, Threads, Pinterest, Bluesky) directly inside the Blotato dashboard under Settings > Accounts. After your accounts are linked, you connect Blotato to Claude Code through an MCP integration. The Blotato MCP setup guide provides the exact configuration for Claude Code in the terminal, VS Code, and other MCP clients. Paste those instructions into Claude Code and it handles the rest. You might need to troubleshoot a few authentication steps, but the integration is straightforward once complete. Blotato also includes a YouTube transcript scraper built into the MCP. That means Claude Code can pull transcripts from any public YouTube video, analyze the content, and write platform-specific social posts from your own original material.Repurposing YouTube Videos Into Social Posts
Here is where the real value kicks in. Instead of watching your own video and manually writing captions, you hand Claude Code a YouTube URL and a skill file. I tested this with a recent podcast episode. The prompt was simple: "Use my social media skill to create five posts for X and LinkedIn based on this YouTube video." Claude Code scraped the transcript through Blotato, referenced my voice and formatting rules from the skill file, and generated all ten posts in a few minutes. The skill file is doing the heavy lifting here. Without it, you get generic AI captions that sound like everyone else on the internet. With a properly built Claude Code Skills Stack, the output matches your actual tone, follows platform character limits (280 for X, 500 for Threads), and structures captions differently for professional platforms like LinkedIn versus short-form platforms like TikTok. From there, posting is instant. Tell Claude Code "post X post 3 to my account right now" and it publishes directly through Blotato. A live link appears in your terminal within seconds. For content you want to schedule, specify a date and time. Claude Code even checks your existing Blotato calendar for conflicts before locking in the slot.
Batch Scheduling Short-Form Video Content
Single posts are useful for quick wins. Batch scheduling is where this system saves hours every week. I keep a Google Drive folder where my short-form video editor drops finished clips. By connecting the Google Workspace CLI to Claude Code, I can point it at that folder and say "write captions for all ten videos and schedule them starting June 8th, Monday through Friday, 8 AM Central." Claude Code opens the folder, scrapes each video transcript, writes two caption groups per video (Group A for Instagram, TikTok, and YouTube Shorts with hashtags; Group B for X and LinkedIn with a different structure), uploads the video files to Blotato, and schedules everything across platforms. The entire batch took a few minutes to process. One practical note on tokens: this type of task burns through usage quickly. I recommend switching to Sonnet 4.6 instead of Opus, setting effort to low, and turning off extended thinking mode. Those three adjustments dramatically reduce token consumption without sacrificing caption quality for social media management tasks.
