How connecting works
The Model Context Protocol (MCP) is an open standard that lets AI assistants use tools from other services. AdVentur.ai runs a remote MCP server, so you connect it by URL. There is nothing to install and no API key to copy.
https://api.adventur.ai/mcpWhat happens when you connect
Section titled “What happens when you connect”- Your client registers itself. It reads the server’s OAuth metadata and registers as a client automatically (RFC 7591 dynamic client registration).
- You sign in and consent. A browser window opens on AdVentur.ai. You sign in, pick a workspace and choose which permissions to grant. The consent page shows the client’s name and where it will send you back, so a look-alike can’t hide.
- The client gets a token. It receives a short-lived access token for that workspace and those permissions, and refreshes it in the background.
- The assistant sees your tools. It lists the tools your permissions allow. Tools you didn’t grant are left out, and the server tells the assistant which ones are hidden so it can ask you rather than guess.
From then on, when you ask for something in chat, the assistant decides which tools to call. Your client may ask you to confirm calls that change things.
Supported clients
Section titled “Supported clients”What an assistant can and can’t do
Section titled “What an assistant can and can’t do”Can, with the permissions you grant: read your brief, strategy, posts and analytics; draft posts, threads, replies, DMs, images and carousels; approve and schedule drafts; publish an approved post now; delete your own X posts; update your brand kit.
Can’t, whatever you grant: connect or disconnect social accounts, change billing, invite or remove team members, change autonomy settings, publish a post that hasn’t been approved, or see other workspaces.