MCP overview
Control AgentReacher from any MCP-compatible AI assistant.
What is MCP?
MCP (Model Context Protocol) is an open protocol that lets AI assistants like Claude, Cursor, or Windsurf interact directly with external tools. AgentReacher implements MCP so you can manage your entire social media workflow from inside your AI assistant. No dashboard needed.
How it works
AgentReacher exposes a set of tools through the MCP protocol. When you connect your MCP client, those tools become available inside your AI assistant. You talk to your assistant in natural language and it calls the right AgentReacher tools on your behalf.
The web dashboard and MCP connect to the same backend. Everything you do through MCP shows up in the dashboard, and vice versa.
What you can ask your AI to do
With MCP connected, your AI assistant can handle your full social media workflow. Here are some examples:
- “Create a post about our new feature for X, LinkedIn, and Instagram”
- “Schedule it for tomorrow at 9am”
- “Show me everything scheduled for this week”
- “Update the draft to use a different headline”
- “Delete the post I created yesterday”
- “What social accounts are connected?”
- “Disconnect Bluesky from this workspace”
- “Invite maria@example.com to the workspace”
- “Switch to my other workspace”
- “Create a new workspace called Summer Campaign”
- “Publish the draft about the conference right now”
Your assistant figures out which tool to call and what parameters to pass. You just describe what you want.
Setting it up
- Install AgentReacher from your MCP client's registry, or run
npx agentreacher@latest initto set up your MCP connection. - On first use, AgentReacher opens a browser so you can sign in and authorize the connection. You do not need to create an API key manually unless you want to.
- If you used the CLI setup, paste the generated configuration into your MCP client's settings. For example, in Claude Desktop add it to
claude_desktop_config.json. - Restart your MCP client. The AgentReacher tools appear in your tool list and your assistant can start using them.
Workspaces and MCP
Your MCP session is scoped to a workspace. You can switch workspaces, create new ones, or delete them, all from your AI assistant. Each workspace has its own connections, posts, and members.