If you haven't connected an AI client to your directory's MCP server yet, this is the fastest path from "I have a directory" to "I have an agent that can actually work on it."
Step 1: Get your organization API key
In your dashboard, go to Settings → Developer API and generate a key. Keep it private — it's scoped to your organization's directories and anything holding it can read and write them.
Step 2: Add the server to Claude Desktop's config
Add an entry pointing at your directory's MCP endpoint with your key as a bearer token in the headers. The exact JSON block is in the MCP integration guide, and it's the same shape for Claude Code or any other MCP-compatible client.
Step 3: Restart and verify
Restart the client and start a new conversation. Ask it to list your directories — if it comes back with real data, you're connected. The server also returns usage instructions on initialize, so most clients will already have a working sense of what's available.
Step 4: Start with something read-only
Before asking for edits, ask for an audit: directory health, claim requests waiting for review, or a summary of recent activity. It's a good way to confirm the connection is scoped correctly and to see what the agent notices before you hand it write access to anything.
Step 5: Move to real work
From there, bulk listing cleanup, category work, claim review, and content drafts are all a conversation away — no separate dashboard tab required, and everything the agent changes shows up in your dashboard exactly as if you'd done it by hand.