Skip to main content

Choosing MCP Servers

MCP servers are curated collections of related tools organized by workflow or service category. This guide helps you find, evaluate, and enable the right MCP servers for your needs.

What Are MCP Servers?​

An MCP server is a pre-configured collection of tools that work together for a specific purpose:

Marketing Automation MCP server might include:

  • Email campaign tools (Smartlead, Resend)
  • CRM tools (HubSpot)
  • Analytics tools (Google Analytics)

Developer Tools MCP server might include:

  • Code hosting (GitHub, GitLab)
  • CI/CD (GitHub Actions)
  • Issue tracking (Linear, Jira)

Content Creation MCP server might include:

  • Image generation (DALL-E, Midjourney)
  • Text tools (OpenAI GPT)
  • Social media (Twitter, LinkedIn)

Discovering MCP servers​

Opening the unified list​

  1. Log in to MCPBundles Dashboard
  2. Click MCP Servers in the left navigation
  3. Use segments All, In workspace, or Directory (plus search and filters)

You'll see workspace servers and directory listings in one list.

MCP server card information​

Each MCP server card shows:

MCP server name - e.g., "Marketing Automation Suite"

Category Badge - e.g., "Marketing", "Developer Tools"

Tool Count - Number of tools included (e.g., "12 tools")

Provider Count - Number of services used (e.g., "4 providers")

Status label (examples):

  • Ready — Connections verified; tools can run
  • Needs setup — Connect providers or add the server to your workspace
  • Needs attention — Reconnect or fix expired / scope issues
  • Disabled — Turned off for your workspace

Description - What the MCP server does and who it's for

Finding the Right MCP Server​

Search by name or tool​

Use the search box at the top to find MCP servers:

"email" → Shows MCP servers with email tools
"github" → Shows MCP servers with GitHub integration
"marketing" → Shows all marketing-related MCP servers

Search matches:

  • MCP server names
  • Tool names
  • Provider names
  • Descriptions

Filter by Category​

Use the category dropdown to narrow results:

Popular Categories:

  • Marketing & Sales - Email, CRM, outreach tools
  • Developer Tools - Code, CI/CD, version control
  • Content Creation - AI generation, design, media
  • Collaboration - Chat, project management, docs
  • Analytics & Insights - Data analysis, reporting
  • E-commerce - Payment, inventory, shipping

All 35+ categories available in the dropdown.

Filter by Status​

Use the status filter to see:

  • All MCP servers - Full unified list
  • Enabled - MCP servers you've activated
  • Disabled - MCP servers you haven't enabled yet

Evaluating an MCP server​

Before enabling an MCP server, understand what's required:

1. Click on an MCP server card​

Opens the MCP server detail page showing:

Overview Section:

  • Full description
  • Use cases
  • Example prompts to try

Tools List:

  • All tools included
  • What each tool does
  • Input parameters required

Providers List:

  • Services the MCP server uses
  • Your connection status for each
  • Which credentials are needed

2. Check provider requirements​

Look at the "Required Providers" section:

Status indicators:

  • Connected - You have valid credentials
  • Not Connected - Need to add credentials
  • Error - Credentials failed validation

Example:

Marketing Automation MCP server requires:
Smartlead (Connected)
Resend (Not Connected)
Google Analytics (Connected)

This MCP server shows "Needs setup" - Smartlead and Google tools work, but Resend tools are hidden until you connect Resend.

3. Review Tool List​

See all tools you'll get access to:

Tool cards show:

  • Tool name (e.g., smartlead_list_campaigns)
  • Description
  • Input parameters
  • Provider it belongs to

Questions to ask:

  • Do I need all these tools?
  • Are there any tools I'll actually use?
  • Would a smaller, focused MCP server be better?
tip

The 50-Tool Problem: Your AI gets confused with too many tools. Choose MCP servers with tools you'll actually use, not "everything just in case."

Connect and use an MCP server​

Step 1: Add to your workspace​

From Directory (or All), open a listing and Connect — or open a server already In workspace. If the server was Disabled, use Add to workspace / enable actions on the detail page until it is active for your workspace.

Header labels (examples):

  • Ready — Required connections verified; you can use Quick Setup
  • Needs setup — Connect providers in Connections
  • Needs attention — Reconnect or fix scope errors
  • Disabled — Not active in this workspace

Step 2: Finish Connections​

  1. Scroll to Connections on the server detail page.
  2. For each provider row, Connect or Reconnect and complete the panel.
  3. Use Validate Now when the connection is not already VERIFIED.

See Setting up credentials and Validating credentials.

Step 3: Add to your AI​

Copy the MCP URL from Quick Setup and follow Connecting your AI.

Managing your MCP servers​

Viewing workspace servers​

Use MCP Servers → In workspace (or All) to see servers you connected, composed private servers, and remote URLs you added.

Turn off a server​

  1. Open the server detail page.
  2. Click Turn off (or remove workspace access when you do not own the server).
  3. Tools stop routing through that server for your workspace; saved connections usually remain for when you turn it back on.

Turn on again​

Find the server under Directory or All and connect / enable it for the workspace again.

MCP server readiness states​

The header uses a small set of labels from the same readiness model everywhere (list, detail page, Studio):

LabelMeaning
ReadyRequired connections verified; open Quick Setup
Needs setupMissing connection(s) on this server
Needs attentionConnection exists but failed validation or scopes
DisabledNot enabled in this workspace

Partial setup means some providers are VERIFIED and others still need Connect — tools for missing providers fail until fixed.

Fix connections appears when several providers need setup and at least one connection is broken — use Connections on that server until the header clears.

Choosing the right MCP servers​

Start Small​

For beginners:

  1. Enable 1-2 MCP servers to start
  2. Focus on your most common tasks
  3. Add more as you get comfortable

Why?

  • Easier to learn what tools do
  • Less overwhelming for your AI
  • Better performance

Use the Hub for Everything​

Instead of configuring multiple MCP server URLs in your AI client, consider using the Hub endpoint. The Hub aggregates all your enabled MCP servers into a single MCP URL with dynamic tool discovery. Enable and disable MCP servers freely — the Hub picks up changes automatically.

Combine Wisely​

Good combinations:

  • Marketing + Analytics
  • Developer Tools + Project Management
  • Content Creation + Social Media

Avoid:

  • Enabling every MCP server "just in case"
  • Overlapping tools (e.g., 3 different email MCP servers)
  • Unrelated workflows (confuses your AI)

Consider Your Workflow​

Questions to ask:

  • What tasks do I do most often?
  • Which services do I already use?
  • What credentials do I already have?
  • What would save me the most time?

Example: If you're a marketer who uses Smartlead and HubSpot daily, start with the "Marketing Automation" MCP server. Add "Analytics" MCP server later if needed.

Custom and composed MCP servers​

Private servers combine included catalog or workspace templates behind one URL (composition), or add an external server with Add remote MCP URL. There is no global per-tool checklist.

Team workspaces share servers you create according to workspace access rules.

Troubleshooting​

MCP server won't enable​

Possible causes:

  1. Backend error

    • Fix: Refresh page and try again
    • If persists, contact support

  2. Permission issue

    • Fix: Make sure you're logged in
    • Check your account status

MCP server stuck on Needs setup​

Cause: Missing connections on this server (common on composed servers if you only connected providers on a catalog copy).

Fix: Open Connections on the server whose MCP URL you use → Connect / Validate Now until the header shows Ready.

Tools not appearing in AI​

  1. Header Ready on the server tied to the URL in your client.
  2. Workspace API key sent on MCP HTTP requests when using hosted URLs (Using API keys).
  3. Connecting your AI steps for your host.
  4. Restart the AI client after URL or connection changes.

Best Practices​

MCP server selection​

DO:

  • Start with 1-2 relevant MCP servers
  • Enable MCP servers for tools you actually use
  • Review tool lists before enabling
  • Connect all providers for full functionality
  • Disable MCP servers you're not using

DON'T:

  • Enable every MCP server available
  • Leave MCP servers in "Needs setup" state without connecting providers
  • Enable overlapping MCP servers with duplicate tools
  • Ignore credential validation warnings

Credential Management​

  • Connect providers on each MCP server URL you actually call
  • Validate all credentials immediately
  • Keep credentials up to date
  • Re-validate if MCP server status changes to "Error"

Performance​

  • Fewer MCP servers = faster AI responses
  • Focus on quality over quantity
  • Disable unused MCP servers to reduce load
  • Create custom MCP servers to minimize tool count

Next Steps​

Need Help?​