Connect Claude, ChatGPT, Grok, Cursor, Codex, or Gemini CLI to Socialync

Draft, schedule, and publish social media posts from Claude, Claude Code, ChatGPT, Grok, Cursor, Codex, Gemini CLI, or any AI assistant that supports the Model Context Protocol. Included on every plan, the free one too.

Read this first: what MCP can do and how you stay in control

You are always the one who decides what goes live. Nothing publishes until you say so, either with a confirmation in chat or by approving it in the Socialync app.

What it can do

Draft, schedule, and publish posts to eight of Socialync's nine platforms: Instagram, TikTok, YouTube, X, Facebook, LinkedIn, Threads, and Bluesky.
Attach media. Up to 10 images as a carousel (from public URLs or images you paste into chat), and one video per scheduled post (from media you already uploaded to Socialync, or a public video URL).
Schedule posts for the future within your plan's window (2 months on free, 12 months on paid), or publish right away.
Read your analytics: views, likes, comments, shares, follower counts, and top-performing posts.
List connections, check quota, and read post history, including which posts succeeded and which failed.
Cancel a scheduled post, publish X Articles, and generate platform-optimized captions with Socialync's AI on paid plans.

How publishing works (two ways)

Draft flow (a hard approval gate). The assistant prepares a draft and you approve it in the Socialync app at your draft inbox. Nothing goes live until you tap Approve.
Direct publish (you confirm in chat). The assistant shows you the exact caption, platforms, media, and timing, and only publishes after you say yes in the conversation. Every action is logged so you can always see what was posted and when.

Current limits

Video is scheduled only. Immediate publishing is text and images. Video posts, including TikTok and YouTube, are scheduled, because platform video processing takes a few minutes. YouTube needs a title.
You cannot paste a video file into chat. The chat protocol cannot carry video files to a connector. Upload the video in Socialync first, then ask the assistant to use it.
Editing an existing scheduled post. The assistant can cancel a scheduled post and create a new one, but cannot edit one in place. Use the Socialync app to edit.
Plan limits still apply. The free plan meters its 5 posts a month through MCP too, with up to 5 AI-created posts a day per brand. Paid plans allow 10 AI-created posts a day per brand and 1 API key per brand, and the platforms enforce their own daily ceilings on top.

Setup: two ways to connect

The endpoint is https://mcp.socialync.io/mcp (Streamable HTTP). Use OAuth where your client supports it; use an API key where it does not. Both reach the same server and the same tools.

OAuth: Claude, Claude Code, Cursor, ChatGPT, Grok, Codex, Gemini CLI

URL only, no headers, no keys. Your client registers itself, opens a browser, and you sign in with your Socialync account and choose which brands the assistant may manage. Revoke it any time from Connected apps.

Claude.ai: Socialync is in Claude’s connector directory. Open Settings, then Connectors, then “Browse connectors”, search Socialync, and click Connect. Adding the URL as a custom connector still works too.

Claude Code:

claude mcp add --transport http socialync https://mcp.socialync.io/mcp

Cursor: Settings, then MCP, then add a server with the JSON under this list, or install the Socialync plugin from the Cursor Marketplace.
ChatGPT: Settings, then Connectors, enable Developer mode, create a connector with the URL, and choose OAuth.
Grok (grok.com): open grok.com/connectors, click New Connector, choose Custom, paste the URL, and sign in to Socialync when Grok asks. Grok signs in with OAuth, so there is no API key to copy. On Grok Business and Enterprise, an admin adds the connector first.

Codex: add the server, then sign in.

codex mcp add socialync --url https://mcp.socialync.io/mcp
codex mcp login socialync

Gemini CLI: add this to ~/.gemini/settings.json, then run /mcp auth socialync in the CLI.

{
  "mcpServers": {
    "socialync": {
      "httpUrl": "https://mcp.socialync.io/mcp"
    }
  }
}

Cursor and other clients that take a JSON config with a URL:

{
  "mcpServers": {
    "socialync": {
      "url": "https://mcp.socialync.io/mcp"
    }
  }
}

OAuth callbacks that are known to work end to end (any https redirect URI plus http://localhost and http://127.0.0.1 on any port is accepted):

https://www.cursor.com/agents/mcp/oauth/callback
http://localhost:8787/callback
https://claude.ai/api/mcp/auth_callback
http://localhost/callback
http://127.0.0.1/callback

API key: headless agents, config files, and clients that cannot complete OAuth

Generate a key on the API Keys page (one per brand, included on every plan; you see it once, so copy it somewhere safe), then send it as a Bearer header.

Open API Keys

Headless agents and scripts: anything that runs without a person at a browser, such as a server-side agent built on the Claude Agent SDK. Send the key in the Authorization header, as in the JSON below. TikTok posts need the person's own choice of who can view them, passed as tiktokPrivacyLevel; from 2026-11-16, Socialync refuses an API key TikTok post without one.
Grok Bot (separate from grok.com): Settings, then Plugins, then add a plugin named Socialync with the URL and the Authorization header from the JSON below.
Claude Desktop (config file): Settings, then Developer, then Edit Config, and paste the second snippet.

{
  "mcpServers": {
    "socialync": {
      "url": "https://mcp.socialync.io/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_SOCIALYNC_API_KEY"
      }
    }
  }
}

{
  "mcpServers": {
    "socialync": {
      "transport": {
        "type": "http",
        "url": "https://mcp.socialync.io/mcp",
        "headers": {
          "Authorization": "Bearer YOUR_SOCIALYNC_API_KEY"
        }
      }
    }
  }
}

Grok Bot runs its OAuth sign in from Cursor's cloud

This is about Grok Bot only; grok.com uses the OAuth steps above. The same URL can connect in the Cursor IDE and fail in Grok Bot, so test in Grok Bot itself under Settings, then Plugins. If OAuth never starts or you see “fetch failed”, use the API key header from option B. Keep the key out of anything you commit to git.

Then try it

Ask your assistant something like:

“Draft a LinkedIn post about consistency in fitness for tomorrow morning at 9am.”

It will create a draft. Open your draft inbox to review and approve it, or confirm in chat and it publishes directly.

Things to try

Sample prompts that show off what the MCP server can do.

“Draft an X post announcing our Q3 product launch and schedule it for next Monday at 10am EST.”

“Show me my top 5 performing posts this month and tell me what they have in common.”

“Which platforms am I connected to right now?”

“Schedule the product demo video from my Socialync library to TikTok and YouTube Shorts for Thursday at 6pm, title it 'Launch day'.”

“Draft the same announcement for LinkedIn and Threads, with a longer caption for LinkedIn and shorter for Threads.”

Where do drafts go?

When the assistant uses the draft flow, the post lands in your draft inbox and expires after 24 hours if you do not approve it. When you ask the assistant to publish directly, it posts after you confirm in chat and no draft is created.

Tools your AI can call (22)

Reference for developers building custom agents. The same list is published as JSON at /.well-known/mcp.

list_profiles

List the Socialync brands this connection is allowed to manage. Call first; every write takes a profileId.

list_connections

List a brand's connected social platforms and whether each one needs a reconnect

list_audiences

List saved audiences a post can be targeted at

list_media

List media already in Socialync you can attach by mediaId (the normal path for video)

create_media_upload

Get presigned upload URLs for images or a video the assistant generated

finalize_media_upload

Finalize an upload into a single mediaId ready to post

check_quota

Check posting quota, plan, per-platform daily limits, and the scheduling window

validate_post

Preflight a post before anything is submitted: caption length, link count, media type, aspect ratio, video duration, and connection health per platform

create_post_draft

Draft a text or image post for immediate publish (you approve in the app)

schedule_post_draft

Draft a post (text, image, or video) for a future time (you approve in the app)

get_draft_status

Check whether a draft is pending, approved, rejected, or expired

approve_draft

Publish a pending draft after the user confirms in chat

publish_now

Publish a text or image post immediately, after in-chat confirmation

create_x_article_draft

Draft a long-form X Article (needs X Premium on the connected account)

publish_x_article_now

Publish an X Article immediately, after in-chat confirmation

publish_scheduled

Schedule a post (text, image, or video, including TikTok and YouTube) directly, after in-chat confirmation

get_scheduled_posts

List pending scheduled posts

delete_scheduled_post

Cancel a scheduled post, or a whole multi-platform group

get_post_history

Read recent post history with per-platform success and failure

get_analytics

Aggregate analytics with period-over-period changes

get_top_posts

Top-performing posts ranked by engagement

generate_content

Generate platform-optimized captions with Socialync's AI (paid plans)

Learn more about MCP·Server docs on GitHub·Listed in the Official MCP Registry

{ "mcpServers": { "socialync": { "transport": { "type": "http", "url": "https://mcp.socialync.io/mcp", "headers": { "Authorization": "Bearer YOUR_SOCIALYNC_API_KEY" } } } } }