Getting started

MCP server

Creator OS runs a hosted MCP server, so any AI app that supports MCP (Claude, ChatGPT, Claude Code, Cursor, Codex and more) can publish, schedule, read analytics and answer comments and DMs for your socials. Nothing to install: add one URL.

https://mcp.creatoros.ca/mcp

Claude and ChatGPT (sign in with Creator OS)

In Claude, open Settings → Connectors → Add custom connector. In ChatGPT, add it as a connector in developer mode. Paste the URL above and connect. You'll sign in to Creator OS, pick the workspace the app gets (agency plans: one client per connection), and choose its access:

Read & write

Everything: publish, schedule, edit, reply, run automations, plus all reads

Read only

Posts, analytics, followers, comments and DMs. It can’t publish, reply or change anything

Each connection gets its own key. See and disconnect them in Settings under Connected apps.

Claude Code, Cursor, Codex (API key)

Tools that let you set headers can skip sign-in and use a workspace's API key from Settings → API keys. Claude Code:

$ claude mcp add --transport http creatoros https://mcp.creatoros.ca/mcp --header "Authorization: Bearer cos_live_..."Copy

Anything that takes an mcp.json (Cursor, Windsurf, VS Code):

{
  "mcpServers": {
    "creatoros": {
      "type": "http",
      "url": "https://mcp.creatoros.ca/mcp",
      "headers": { "Authorization": "Bearer cos_live_..." }
    }
  }
}

Want an agent that can look but not touch? Connect through sign-in and choose read only. A read-only connection only sees the read tools, and the API refuses writes from it.

Tools

get_workspace, list_accounts, get_account_health, get_follower_stats

Who you are connected as, your connected socials, and follower growth

list_posts, get_post

The content calendar: drafts, scheduled, published and failed posts

get_post_analytics, get_daily_metrics, get_best_time_to_post

Views and engagement per post, daily totals, and when your audience is most active

list_posts_with_comments, list_comments, list_conversations, list_messages

Comments and DMs across every network

list_automations, get_connect_link, check_caption_length

Comment-to-DM automations, a link to connect a new account, and caption limits

list_webhooks, list_webhook_deliveries

Your webhook endpoints and their delivery log

skool_list_communities, skool_list_posts, skool_find_posts, skool_get_comments

Your connected Skool communities, their feeds and comment threads

skool_list_members, skool_pending_members, skool_member_emails

Skool members (active, churned, banned…), join requests and collected emails

skool_profile, skool_followers, skool_list_chats, skool_read_chat

Your Skool follower count, followers list and DMs

skool_revenue, skool_growth, skool_chart, skool_engagement

Skool MRR, 30-day growth, dashboard charts and active members

skool_create_post, skool_multi_post, skool_comment, skool_update_post, skool_pin_post, skool_like_post, skool_approve_member

Post (to one or several communities), reply, edit, pin, like and approve in Skool (read & write)

skool_send_message, skool_start_chat, skool_block_user

Send Skool DMs, start conversations, block or unblock (read & write)

create_post, update_post, upload_media_from_url

Publish, schedule or draft posts (read & write)

reply_to_comment, hide_comment, send_message

Answer comments and DMs (read & write)

create_automation, update_automation

Set up keyword comment-to-DM funnels (read & write)

create_webhook, test_webhook

Register an endpoint for events and send it a test (read & write)

delete_post, delete_comment, delete_automation, delete_webhook, disconnect_account, skool_delete_post, skool_deny_member

Destructive: marked so your AI app asks you before running them (read & write)

Every tool runs through the Creator OS API, so ids, errors and limits match the API reference. Media for posts can be any public image or video URL, or a med_ id from upload_media_from_url.