Here is a scenario you probably know: you spend three hours writing a blog post, publish it, and then realise you need LinkedIn posts, an email version, and social snippets for every channel. So you spend another hour manually rewriting paragraphs, copying quotes, reformatting for each platform, and pasting everything into Buffer. By the time you are done, you have spent half your day distributing one piece of content.
Content repurposing saves 60 to 80 percent of creation time compared to building each piece from scratch, according to a 2026 analysis by BlogHunter. Teams that repurpose systematically see a 40 percent increase in overall content output without growing their team. Yet only 35 percent of marketers actively repurpose content across channels, according to HubSpot's 2026 State of Marketing report. The gap is not awareness. It is that the manual process takes too long.
We at Supernodes build systems that automate this kind of repetitive work. The workflow we use internally, and the one we walk through here, connects Claude Cowork to Buffer via its MCP server. Once it is set up, you paste a blog draft and get ten pieces of scheduled content back, ready to publish. And because of the way tools like Claude Cowork and Buffer's API have matured in 2026, anyone with a Claude Pro subscription and a free Buffer account can set it up in about 30 minutes.
What you need before starting
You need three things. A Claude Pro account ($20 per month) with the Claude Desktop app installed. A Buffer account, which is free for up to three channels. And Node.js version 18 or higher, which you can download for free from nodejs.org. That is the full list. No API keys to manage, no custom code to write, no separate hosting.
Buffer's API, which includes its MCP server, is available on every plan including free. The API gives you up to 3,000 requests per month on the free plan, which covers daily repurposing for a single site easily. Buffer supports LinkedIn, X, Instagram, Threads, Facebook, TikTok, Pinterest, YouTube, Bluesky, Mastodon, and Google Business Profile through the same MCP connection. For more on how MCP connects agents to infrastructure, read our guide to the 10-layer AI agent stack.
Step 1: Connect the Buffer MCP to Claude
Set up the Buffer MCP server in Claude Desktop
Open Claude Desktop and go to the Customize tab. Select Connectors, hit the plus button, then choose Add custom connector. Paste this URL: https://mcp.buffer.com/mcp. Name it "Buffer". Claude will prompt you to log into your Buffer account and authorise the connection.
If you prefer more control, you can connect via the Buffer API directly. Go to Buffer's Integrations settings, pick Claude, and follow the prompts to generate an API key. Then paste the configuration snippet into Claude Desktop's Settings, Developer, Edit Config and restart Claude. The result is the same: Claude can now read your Buffer queue, draft posts, schedule them, and edit existing content.
node -v in your terminal.
Step 2: Write a brand voice Skill for Claude
This step matters more than the tool setup. Claude can repurpose content without a brand voice file, but the output will be generic. A brand voice Skill is a text file where you write your rules in plain English. No code, no special syntax. If you can write a Notion page, you can write a Skill.
Create a folder called .claude/skills/ in your Claude project directory. Inside it, create a file called brand-voice.md with your tone guidelines, banned phrases, and audience description. According to practitioners who use Claude Cowork for marketing tasks, this takes about 30 to 45 minutes to write the first time. Once it exists, Claude loads it automatically on every task. Every piece of content it produces follows the same rules without you reminding it.
Step 3: Give Claude the blog draft and the output spec
Open Claude Cowork, not the regular chat, and paste your blog post. Then tell it what you want back. Be specific about the formats and platforms.
Example prompt for Claude Cowork:
"Turn this blog post into the following pieces: 3 LinkedIn posts (problem, solution, stat), 2 email snippets (subscribers, cold audience), 1 Twitter thread (5 tweets), 2 Instagram captions (educational, hook-driven), 1 newsletter intro, 1 key takeaway summary. Use the brand voice Skill. Review each piece against it before outputting."
Claude Cowork reads the blog post, applies the brand voice Skill, generates each format, and gives you a complete document. The key difference from just asking a chatbot to do this: Cowork reads your actual content, follows your specific rules, and produces drafts that match the tone of what is already working on your channels. It can also browse your LinkedIn page to check what types of posts are getting engagement before drafting, which changes the output quality noticeably. We cover the same search-and-synthesise pattern in our guide to building deep research AI agents.
Step 4: Review and schedule in one prompt
Schedule to Buffer
Tell Claude: "Schedule each piece to the right channel. Post the three LinkedIn pieces over Monday, Wednesday, and Friday. Put the Twitter thread on Tuesday morning." Claude pushes each post to the correct channel with the correct timing.
If you need to edit a post that is already scheduled, you can do that too. "Change the first LinkedIn post. Update the opening line to this version instead." Claude finds the post in your Buffer queue, makes the edit, and saves it. No logging into Buffer, no hunting through the calendar.
Step 5: Automate the whole thing with /schedule
Set up a recurring task
In Claude Cowork, type: "/schedule Check my blog folder for new drafts every Thursday at 3pm. If there is a new draft, repurpose it using the same format as my last run and schedule the pieces to Buffer." Claude will ask you to confirm the schedule parameters. Once confirmed, it runs autonomously. By Friday morning, your social queue is full for the next week.
What ten pieces look like from one post
Here is the output breakdown from a single 1,500-word blog post. Three LinkedIn posts target different segments: one frames the problem, one presents the solution, one leads with a surprising stat. Two email snippets cover your subscriber list and cold outreach separately. One Twitter thread with five self-contained tweets, two Instagram captions (educational and hook-driven), one newsletter intro paragraph, and one key takeaway summary with three bullet points. Ten pieces from one post. And because the brand voice Skill applies to everything, the tone is consistent across every format.
Why this beats the manual approach
A 1,500-word blog post takes about three hours to research and write. Repurposing it manually takes another 60 to 90 minutes. With this pipeline, that step takes about seven minutes. According to a 2026 analysis by AutoFaceless, AI-driven content repurposing reduces production costs by up to 65 percent.
And Gartner's May 2026 survey of 402 CMOs found that marketing leaders expect AI-driven automation of marketing work to double from 16 percent to 36 percent by 2028. The teams that already have these workflows running will be the ones defining what that looks like rather than catching up to it.
Frequently asked questions
What is the Buffer MCP and how does it work with Claude?
The Buffer MCP is a Model Context Protocol server that connects AI assistants like Claude to your Buffer account. Once connected, Claude can see what is scheduled, draft new posts, and schedule them directly.
How much does it cost?
Claude Pro costs $20 per month. Buffer is free for up to three channels. The Buffer API is included on every plan. Total: roughly $20 to $26 per month, or less if you already subscribe to either tool.
Can I do this with ChatGPT instead of Claude?
Yes. Buffer supports MCP connections for ChatGPT, Cursor, Raycast, and Perplexity. The workflow pattern is identical.
Does content repurposing actually save time?
Multiple studies show repurposing saves 60 to 80 percent of creation time compared to starting from scratch. Only 35 percent of marketers actively repurpose, so there is a significant advantage for those who do.