---
description: Choose, connect and scope marketing MCP servers read-only, test each one and return a stack inventory
argument-hint: [tools and clients or audit what is connected]
---

Use the marketing-mcp-setup skill for this task: $ARGUMENTS

Read martech-context/ first if it exists, especially stack.md, decisions.md and metrics.md.
If $ARGUMENTS is empty or vague, ask for the jobs the connections must do with one known answer each, the tools, accounts and regions in scope, the clients each person uses, and who administers each tool.
Run the skill's config scanner (scripts/mcp_inventory.py from the skill folder with --discover --project <repository>; for a machine audit add --all-projects) before connecting anything new. Never open, print or paste a config file, secret, token or key.
Follow the skill's procedure in order: one server at a time, the strongest read-only layer first, then tests T1 to T6 in every client that uses the server. Finish with its acceptance checks.
Ask before changing any client configuration. Apply it with the client's own command, or hand the user the block and file path to paste; never open the config file. The user completes every sign-in and consent screen. Draft admin changes for the admin instead of making them.
Never test with a delete, send, launch, publish or spend change. The only write attempts are harmless ones expected to be denied, in a sandbox the user names, each exact call approved first.
Hand back the stack inventory, the connection test log, the scanner findings and the proposed martech-context updates: Agent access lines in stack.md, decisions.md entries, mcp-inventory.md, mcp-tools/ snapshots and a README.md line naming them.

If the marketing-mcp-setup skill is not installed, install it with: npx skills add https://www.martechdevs.com --skill marketing-mcp-setup -a claude-code
