Connect Buffer's MCP to Claude, draft and schedule social posts across 11 channels from one chat, then review everything in your queue before it goes live.
Last verified · 2026-08-12 · by Moe Ameen
You can now draft a social post inside a Claude conversation and drop it straight into a scheduling queue without leaving the chat. It works through the Buffer MCP server — Model Context Protocol, the open standard that lets an AI assistant like Claude connect to an outside app with your permission. Claude does the writing and the queue commands; Buffer holds your connected channels and does the actual publishing.
The important thing to get straight before you set it up: Claude is not posting directly to Instagram or LinkedIn. It hands a finished draft to Buffer, where the post lands as a draft or a scheduled item you can see and edit before anything goes live. That review step is the safety net, and it means the setup is genuinely useful for drafting at conversational speed — but it also has hard edges. Claude writes text; it does not make the video, the carousel, or the image, and the connection only runs while you are in a chat with it. This guide walks the full setup and then the honest limits.
The Claude-plus-Buffer setup is fast for the half of the job that is words. Its two hard edges are the same two things it can't touch: it makes no media, and it only runs while you are typing in a chat. Kompozy is built to close exactly those two gaps — it generates the assets Claude can't, and it runs when you're not there.
Start with the media gap. Where Claude hands Buffer text and you go make the visuals yourself, Kompozy generates the visual formats natively from your source: [avatar Persona Shorts](/glossary/persona-shorts) and captioned clips, [Carousel Posts](/glossary/hyperframes) rendered to your exact brand template, Photo Posts, Quote Graphics, Listicle and Naturalistic videos — 18 output formats in all, across video, image, and text. So a single blog or transcript becomes a week of finished, on-brand posts with the pictures already attached, not a stack of drafts waiting on a separate editing pass. Then the automation gap: the Buffer connector only fires inside a conversation you start, while Kompozy's [Autopilot](/glossary/autopilot) ingests a source, generates the batch, and routes it into a per-post review queue on a recurring schedule — the event-triggered publishing the MCP path explicitly doesn't do. Voice is handled the same way you'd hand-tune a Claude prompt, but once: a [Persona Brief](/glossary/persona-brief) pins your positioning, phrasing, and banned words across every generation instead of being re-pasted each session. And publishing is native — it fans the batch across eight social platforms plus blog and email from one queue, with the same review-before-live checkpoint you get in Buffer.
The honest split: if you want to draft a few posts conversationally and you already own your visuals, Claude and Buffer are a clean, cheap way to do it — see [how to cross-post across platforms](/how-to/cross-post-to-all-platforms) for the manual fanout. If the bottleneck is making the video and images and keeping it running without you, that's the line Kompozy is built for. Starter runs $99/mo (5,500 credits); Pro is $299/mo (18,000 credits); Enterprise is custom. Same conversational speed on the drafting, but the engine also makes the assets and does the unattended part.
No. Claude drafts the post and sends it to Buffer through the MCP connection; Buffer is what publishes to the channels you connected there. The post lands in your Buffer queue as a draft or scheduled item, and you review it before it goes live.
MCP access is included on every Buffer plan, including Free, so you don't pay Buffer extra for it. You do need a Claude account, and a paid Claude plan if you want the higher usage limits; those are billed separately by Anthropic.
The channels Buffer supports: LinkedIn, X, Instagram, Threads, Facebook, TikTok, Pinterest, YouTube, Bluesky, Mastodon, and Google Business Profile. You connect each account in Buffer first, then Claude can queue to it.
No. The connection only runs inside a conversation you start, and finished posts sit in your Buffer queue until their scheduled time. There's no autonomous, event-triggered publishing — it's an interactive drafting-and-scheduling workflow, not a hands-off engine.
Not through this integration. Claude writes the text and handles the queue commands; it does not generate visual media here, and media attaches to a queued post only as a hosted URL. You produce the video, carousel, or image separately and add it in Buffer.
Not for the connector path — you add Buffer in Claude's Connectors settings and authorize it, no code required. Buffer also documents a manual API-key setup that runs a local server with Node.js 18+, but that's optional.