Skip to content
MCP

Connect Product Metrics MCP from a coding agent

Add the Product Metrics MCP server to Claude Code, Codex, Cursor or VS Code from the terminal, then give the agent the skills, the productmetrics CLI and a starter prompt.

Add the server

Server URL (remote HTTP; the client starts the browser sign-in): https://www.productmetrics.io/mcp

  • Claude Code: claude mcp add --transport http product-metrics https://www.productmetrics.io/mcp, then sign in through /mcp
  • Codex: codex mcp add product-metrics --url https://www.productmetrics.io/mcp, then codex mcp login product-metrics
  • Cursor or any JSON-config client: {"mcpServers": {"product-metrics": {"url": "https://www.productmetrics.io/mcp"}}}
  • VS Code (.vscode/mcp.json): {"servers": {"product-metrics": {"type": "http", "url": "https://www.productmetrics.io/mcp"}}}

Claude and ChatGPT add the server in the browser instead: see For humans.

A person finishes the sign-in once, in a browser: the agent can’t, and there is no API key or header to set. One connection covers every workspace you can open and lasts 30 days from approval.

https://www.productmetrics.io/mcp is a server URL for your MCP client, not a page to open in a browser.

productmetrics mcp-config prints the URL and the steps for one client, or for all of them without --client:

npm install -g @productmetrics/cli
productmetrics mcp-config --client claude   # or chatgpt, cursor

It prints https://prod-backend.productmetrics.io/mcp, the same server on the API’s own origin. Either URL works.

Verify your connection

Open your client’s MCP list (/mcp in Claude Code) and check that product-metrics is connected. Then try:

Which Product Metrics stores can I open, and what is left to set up?

The agent calls list_properties, then execute with GET /setup/status. If no Product Metrics tools appear, restart or reload the client after saving its MCP configuration.

Give the agent skills

Skills are task instructions an agent loads when it needs them: which tools to call, in which order, and what to check. From the project where the agent works:

npx skills add https://www.productmetrics.io/docs \
  --skill '*' \
  --agent claude-code \
  -y

Use --agent codex or --agent cursor for those agents, and add --global for every project. This installs set-up-tracking, connect-google-ads, check-tracking, bid-on-profit and analyse-google-ads. Needs Node.js 22.20 or later. See Install skills.

The index for agents is at https://www.productmetrics.io/docs/.well-known/agent-skills/index.json.

Give the agent the CLI

productmetrics calls the same API as the MCP server’s search and execute tools. Every API operation is a command, productmetrics <group> <command>, and its flags are the operation’s fields.

# install (Node.js 22 or later), or run once with npx @productmetrics/cli
npm install -g @productmetrics/cli

# sign in in the browser: run this yourself, not through the agent
productmetrics login

# the short guide for AI agents
productmetrics help agents

# where setup stands, and the next step
productmetrics setup get-status --company-id <company_id> --json

What an agent needs to know:

  • Pass --json. stdout then holds exactly one JSON document. --select '.data.properties[].property_id' prints only part of it.
  • No prompts. Without a terminal, or with PM_AGENT=1, CI or --no-input, productmetrics never prompts, never opens a browser and prints no colours.
  • Discover, never guess. productmetrics <group> <command> --help --json gives the exact flags, types and allowed values. Ids come from earlier results.
  • Retries are safe where --help lists --auto-idempotency-key. After any other failed write, read the state before retrying. --dry-run prints the request without sending it.
  • Deletes need --yes. Ask the person first.
  • Exit code 4 means a person must finish a step in a browser: the URL is in next_action.url.

When productmetrics status --json shows signed_in: false, the agent asks you to run productmetrics login in your own terminal. In CI, set PM_TOKEN. More in CLI for agents and JSON and exit codes.

Search the docs

The docs MCP lets the agent search these docs while it works. It reads documentation only: for your own data, use the server above.

claude mcp add --transport http product-metrics-docs \
  https://www.productmetrics.io/docs/api/mcp

Other clients take https://www.productmetrics.io/docs/api/mcp the same way as the Product Metrics server. https://www.productmetrics.io/docs/llms.txt lists every docs page with a one-line description: point the agent at it before it reads further.

Starter prompt

Paste this into your coding agent to set up Product Metrics:

You are setting up Product Metrics for my store.

- Read https://www.productmetrics.io/docs/llms.txt (or use the
  product-metrics-docs MCP) before you act.
- Use the Product Metrics MCP server if it is connected. Otherwise use
  the productmetrics CLI with --json: run `productmetrics help agents`
  first, then `productmetrics status --json`. If I am not signed in,
  ask me to run `productmetrics login` in my own terminal.
- Read the setup state with
  `productmetrics setup get-status --company-id <id> --json` and follow
  hub.company.next_action and hub.property.next_action, one step
  at a time.
- Never guess a command or flag: run
  `productmetrics <group> <command> --help --json`. Ids come from earlier
  results.
- Exit code 4 means I must finish a step: give me next_action.url
  (or the store setup token) and wait for me.
- Pass --auto-idempotency-key where --help lists it. Ask me before
  anything that needs --yes.
- Never ask me for a password or an API key; ask me for store
  credentials only when a step needs them.

More