You have done the basic setup. Smartlead MCP is connected, Claude is pulling live campaign data, and everything is working. Now comes the part most guides skip - what happens when your team wants access too, when you need to run multiple MCP servers at once, or when something breaks and you have no idea why.

This guide covers all of that. If you have not done the initial setup yet, start with the Smartlead MCP setup guide first and come back here once you are connected.

TLDR

What this covers: Advanced Smartlead MCP configuration - running multiple servers simultaneously, setting up access for a team, keeping API keys secure, and troubleshooting the issues that come up after the basic setup is already working.

Three things every sales team needs to know:

• Multiple MCP servers run simultaneously with no conflict - just define each one inside the same mcpServers block
• Team setup means one config per machine and one API key per person - no shared credentials
• You can keep keys out of the config file entirely using environment variables

Verify any config change with three steps: fully restart Claude Desktop, confirm Smartlead appears under the tools indicator, then run a live test prompt.

Running Multiple MCP Servers Simultaneously

The first question most teams ask after getting Smartlead MCP working is whether they can run other MCP servers at the same time - Notion, Slack, a CRM, a browser tool. You can, and the setup is simple.

The claude_desktop_config.json file supports multiple MCP servers inside a single mcpServers object. Each server is a named block and Claude Desktop loads all of them at startup, so you have access to all their tools in every session without switching anything.

Here is what a configuration with Smartlead, Notion, and Slack looks like:

json

{
 "mcpServers": {
   "smartlead": {
     "command": "npx",
     "args": ["-y", "@smartlead/mcp-server"],
     "env": {
       "SMARTLEAD_API_KEY": "your-smartlead-key-here"
     }
   },
   "notion": {
     "command": "npx",
     "args": ["-y", "@notionhq/notion-mcp-server"],
     "env": {
       "NOTION_API_TOKEN": "your-notion-key-here"
     }
   },
   "slack": {
     "command": "npx",
     "args": ["-y", "@modelcontextprotocol/server-slack"],
     "env": {
       "SLACK_BOT_TOKEN": "your-slack-token-here"
     }
   }
 }
}

The critical rule here: every server must be defined inside the same mcpServers object. If you paste a second "mcpServers": { ... } block anywhere else in the file, Claude Desktop will silently use one and ignore the other - no error message, no warning, just missing tools. This is the single most common mistake when adding a second server.

Each named block is independent, since if one server fails to start, the others still load. Claude will tell you which tools are unavailable if a server does not connect. There is no practical limit on the number of servers you can define, though tool counts add up quickly. If you notice slower initial responses with many servers active, remove ones you are not actively using.

Try Smartlead Free and Connect MCP in 5 Minutes

Start your free Smartlead trial and have your first agentic outbound workflow running today.

Team Configuration - How to Set This Up for Multiple Users

Smartlead MCP does not have centralized team authentication. There is no single shared config or admin-distributed credential. The model is one config per machine and one API key per person - and for most sales teams, this is the right setup anyway since it keeps activity traceable and access revocable.

Here is the practical workflow for getting your team set up.

Step 1: Each person generates their own Smartlead API key. In Smartlead, go to Settings then Integrations then API Keys. Create a new key and label it clearly - something like "Claude MCP - [name]" - so you can identify it later.

Step 2: Each person adds the key to their own config file. The location varies by operating system.

• Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/Claude/claude_desktop_config.json

The JSON structure is identical for everyone - only the API key value changes per person:

json

{
 "mcpServers": {
   "smartlead": {
     "command": "npx",
     "args": ["-y", "@smartlead/mcp-server"],
     "env": {
       "SMARTLEAD_API_KEY": "their-individual-key"
     }
   }
 }
}

Step 3: Each person restarts Claude Desktop after editing the config. Changes do not take effect until Claude Desktop is fully quit and reopened - not just the chat window but the application itself. On Mac, quit from the menu bar icon.

Why separate keys matter in practice: API activity in Smartlead is logged per key. If Claude makes an unexpected change to campaign settings, or someone's session triggers a rate limit, you can trace exactly which key was responsible. You can also revoke one person's key without affecting anyone else. Shared keys make both of those things impossible and if someone leaves the team or you suspect a key has been exposed, you regenerate just that one key in Settings then Integrations then API Keys without touching the rest of the team's setup.

When To Use Environment Variables Instead of Hardcoding Keys

If you are working in a shared environment, version-controlling your dotfiles, or following stricter security policies at your org, you can reference environment variables in the config file instead of pasting the API key directly.

Claude Desktop supports environment variable expansion in the env block:

json

{
 "mcpServers": {
   "smartlead": {
     "command": "npx",
     "args": ["-y", "@smartlead/mcp-server"],
     "env": {
       "SMARTLEAD_API_KEY": "${SMARTLEAD_API_KEY}"
     }
   }
 }
}

The syntax is ${VARIABLE_NAME} and Claude Desktop resolves this at startup by reading the variable from your shell environment.

On Mac or Linux, add this line to your shell profile - ~/.zshrc for Zsh or ~/.bashrc for Bash:

bash

export SMARTLEAD_API_KEY="your-key-here"

Then reload the profile:

bash

source ~/.zshrc

On Windows, set the variable through System Properties then Environment Variables, or use PowerShell:

powershell

[System.Environment]::SetEnvironmentVariable("SMARTLEAD_API_KEY", "your-key-here", "User")

After setting the variable, restart Claude Desktop completely so it picks up the new environment.

When is this worth doing? If you commit your dotfiles or config directory to a Git repository, environment variable expansion keeps the actual key out of version control. The config file itself contains no sensitive data - just the variable name. The key lives in your shell profile which you keep private. For most individual setups where the config file stays on your local machine, hardcoding the key directly is fine. The environment variable approach is mainly useful when the config file itself might be shared, synced, or accidentally exposed.

API Key Security - What To Actually Do

The claude_desktop_config.json file works like a .env file - it contains credentials that give programmatic access to your Smartlead account. Treat it the same way you would any credentials file.

Do not commit the config file to version control. If you track dotfiles in a Git repository, add it to your .gitignore:

# .gitignore
claude_desktop_config.json
Library/Application Support/Claude/claude_desktop_config.json

Regenerate keys immediately if you suspect exposure. If your config file was shared, synced, or accidentally committed, go to Settings then Integrations then API Keys in Smartlead and regenerate the key. The old key is immediately invalidated. Update your config file with the new key and restart Claude Desktop.

Audit key activity periodically. In Smartlead's API Keys settings, you can see when a key was last used. If you see activity you do not recognise, revoke and regenerate straight away.

Troubleshooting Beyond JSON Errors

The basic setup guide covers the most common issue, which is malformed JSON in the config file. Here are the problems that come up after the initial setup is already working.

Tools appear but return empty data

This usually means the API key is valid but does not have access to the workspace you are querying. On team accounts, different API keys may have access to different workspaces or subaccounts. Check in Smartlead that the key you are using belongs to an account with access to the campaigns, leads, and data you are asking about.

Intermittent tool failures mid-session

The most likely cause is Smartlead's API rate limits. If you run a complex prompt that triggers many sequential tool calls - pulling campaign lists, then lead counts, then statistics for each campaign - you may hit rate limits since Claude does not automatically retry throttled requests.

The practical fix: break multi-part questions into smaller requests. Instead of give me a complete breakdown of all my campaigns with open rates and reply rates for the last 30 days, ask for one campaign or one metric at a time. This spreads the API calls out and stays within limits.

Tools disappear after working fine

This almost always resolves with a complete Claude Desktop restart. Closing the window on Mac often leaves the application running in the background. To fully restart - look for the Claude icon in the Mac menu bar, click it, and choose Quit Claude. Then reopen the application since MCP servers initialise fresh at each startup.

If a full restart does not fix it, open the config file and check for recent edits. A syntax error introduced since the last working session - even a single missing comma - will prevent MCP servers from loading.

MCP servers never appear

A valid JSON file that still fails to load servers usually has one of two causes. Either the file is at the wrong path - double-check the exact location for your OS - or npx is not accessible. Claude Desktop runs npx to start MCP servers, and if npx is not in your system PATH the servers will not start. Run which npx in your terminal. If it returns nothing, install Node.js from nodejs.org since it includes npx.

How To Verify Your Configuration Is Working

After any configuration change - adding a server, updating a key, switching from hardcoded to environment variable - run these three checks before moving on.

Step 1: Restart Claude Desktop completely. Not just the window - quit the application and reopen it.

Step 2: Look for the tools indicator. In the Claude Desktop chat interface, look for the hammer icon in the input area and click it. You should see a list of connected MCP servers. If you expand Smartlead's entry, its tools should appear. If Smartlead does not appear in the list, the server did not connect - go back and check the config file for syntax errors.

Step 3: Run a live test prompt. Ask Claude How many Smartlead campaigns do I currently have active? This requires a live API call. If it returns real data from your account - You have 4 active campaigns - the configuration is working end to end. If it returns a generic response or says it cannot access Smartlead, the connection failed despite the tools appearing in the list.

A successful response to Step 3 means your API key is valid, the MCP server is running, Claude can call it, and the tool is returning live data from your Smartlead workspace. You are good to go.

The Bottom Line

Getting Smartlead MCP working is the five-minute part. Getting it working reliably across a team, keeping credentials secure, and knowing exactly what to check when something breaks - that is what separates a setup that lasts from one that causes headaches two weeks in.

The patterns here are not complicated. One key per person, all servers in one block, a full restart after every config change, and a live test prompt to confirm everything is actually connected.

Get those habits right from the start and you will spend a lot less time debugging and a lot more time using Claude to run your outbound diagnostics the way it was meant to work.

Get started with Smartlead for free and set up MCP for your team today.

FAQs

Can I run multiple MCP servers at the same time in Claude Desktop?

Yes. Add each server as a named block inside the same mcpServers object in your claude_desktop_config.json. All servers defined in the file load simultaneously when Claude Desktop starts and Claude has access to all their tools in every session.

How do I set up Smartlead MCP for my sales team?

Each team member needs their own Claude Desktop installation and their own Smartlead API key. The config structure is identical for everyone - the only difference is each person's individual key. Generate keys per person in Settings then Integrations then API Keys in Smartlead. This allows per-key activity logging and lets you revoke one key without affecting anyone else.

Is it safe to put my Smartlead API key in the config file?

The config file on your local machine is not publicly accessible, so hardcoding the key is generally fine for individual use. The risk is accidental exposure - committing the file to a repository, syncing it to a shared drive, or screensharing while it is visible. To avoid this, use environment variable expansion in the config and set the actual key in your shell profile instead.

What happens if I have two mcpServers blocks in my config?

Claude Desktop silently uses one and ignores the other - no error, no indication in the UI. The valid structure has exactly one mcpServers object containing all server definitions. If you add a second server and some tools stop appearing, this is the first thing to check.

Can I use environment variables instead of hardcoding my API key?

Yes. In the env block of your config, use the syntax "SMARTLEAD_API_KEY": "${SMARTLEAD_API_KEY}". Claude Desktop resolves the variable from your shell environment at startup. Set the variable in your .zshrc or .bashrc on Mac/Linux, or in System Environment Variables on Windows, then restart Claude Desktop after setting it.

How do I know if my Smartlead MCP configuration is working?

Three checks: fully restart Claude Desktop, verify Smartlead appears under the hammer icon in the chat interface, then ask "How many Smartlead campaigns do I currently have active?" If Claude returns real data from your account, the connection is working end to end.