Skip to main content

​
What is Runlayer?

Runlayer is a secure MCP platform that lets you connect AI clients (like Claude, ChatGPT, Cursor, and more) to the tools and resources you use like Github, Atlassian, DataDog, Hubspot and 18k+ more. Think of MCP servers as plugins or extensions that give AI clients new skills like opening a PR in GitHub, pulling a ticket from Jira or researching about a deal from Salesforce. Key Benefits:
  • One-Click Setup: Install and connect MCPs to any AI client instantly
  • Team-Wide Access: Approved MCPs are automatically shared across your organization
  • Cross-Platform: Configure once and use the same MCP in Claude, Cursor, ChatGPT, and more
  • Full Visibility: All MCP activity is logged, secured, and controlled by your admins

​
Getting Started

​
First Login

Sign In
  1. Navigate to your organization’s Runlayer URL (provided by your admin)
  2. Click “Sign In” - You’ll be redirected to your organization’s Single Sign-On (SSO) page
  3. Log in with your work credentials - Same username and password you use for other company tools
MCPs After logging in, you’ll see the main dashboard with these key sections:
  • MCPs: View and manage your MCP servers
  • Catalog: Browse available MCPs you can add
  • Audit Logs: View activity history
  • Settings: Manage your profile and preferences
Admin users also see Analytics and additional management options. If you need admin access, contact your IT department.

​
Understanding MCP Servers

What is an MCP Server? An MCP server is like a plugin that extends what AI assistants can do. For example:
  • GitHub MCP: Lets AI read and manage your code repositories
  • Google Drive MCP: Gives AI access to your documents
  • Slack MCP: Allows AI to read messages and post updates
  • Database MCP: Connects AI to your company’s databases
Key Concepts:
  • Server: The MCP plugin itself (e.g., “GitHub Integration”)
  • Tools: Actions the MCP can perform (e.g., “create issue”, “search code”)
  • Resources: Data the MCP can access (e.g., repository files, issues)
  • Status:
    • Active - Ready to use
    • Pending - Waiting for admin approval
    • Inactive - Disabled or not configured
    • Rejected - Not approved for use

​
Browsing the MCP Catalog

Catalog The Catalog shows pre-vetted MCP servers you can request to use:
  1. Navigate to Catalog in the left sidebar
  2. Browse or search for MCPs by name or category
  3. Click on any MCP to see details:
    • What it does
    • What tools it provides
    • What data it can access
    • Security considerations
  4. Click “Add Server” to request access (see section below)

​
Connecting MCPs to Your AI Tools

You can connect existing MCPs to Claude, ChatGPT, Cursor, or other AI tools. MCP Clients

​
Installation Steps

1

Open the MCP Details

Click on any Active MCP in your MCPs list
2

Select Your AI Tool

Choose your client from the dropdown (see supported clients below)
3

Complete Setup

Follow the instructions to complete the setup in your AI tool

​
Supported AI Tools & Clients

Runlayer provides tailored setup instructions for each of these clients: IDEs & Code Editors:
  • Cursor - One-click connect button
  • VS Code - One-click connect button
  • Windsurf - Configuration file setup
  • GitHub Copilot - Configuration file setup
AI Assistants:
  • Claude Desktop - Configuration file setup
  • Claude Code - CLI command setup
  • ChatGPT - Web UI connector setup
  • Codex - Proxy URL and API key setup
Terminal & CLI Tools:
  • Warp - Configuration file setup
  • Gemini CLI - Configuration file setup
Automation & Integration:
  • Gumloop - Proxy URL and API key setup
  • Raycast - Configuration file setup
SDKs (for developers):
  • TypeScript SDK - Code-based integration
  • Python SDK - Code-based integration
  • Instructions for any MCP-compatible client
Security Note: Your API key provides access to the MCP server. Never share your API key or configuration files publicly.

​
Requesting a New MCP Server

When you need an MCP that you don’t currently have access to, you’ll need to submit a request to your administrator for approval.

​
When to Request an MCP

Request a new MCP server when:
  • You need capabilities not available in your current MCPs
  • You want to connect AI to a specific data source or API
  • Your workflow requires automation that an MCP can provide
  • You’re following documentation that requires a specific MCP

​
Before You Request

1

Check Existing MCPs

Browse the MCPs page to see if what you need already exists. You might just need access to an existing server.
2

Review the Catalog

Check if your desired MCP is in the pre-approved catalog (faster approval)
3

Prepare Justification

Be ready to explain:
  • What problem you’re solving
  • Why this specific MCP is needed
  • What data it will access
  • Expected business value
4

Gather Technical Details

If requesting a custom MCP, collect:
  • Server URL or connection details
  • Documentation link
  • Authentication requirements
  • Any configuration settings
Add Server

​
Create From the Catalog

Best for: Standard, pre-vetted MCPs like GitHub, Slack, PostgreSQL, etc.
1

Browse the Catalog

Navigate to the Catalog page and find the MCP you need
2

Click 'Add Server'

On the MCP details page, click the “Add Server” button
3

Configure Settings

Fill in any required configuration:
  • API keys for external services
  • Database connection strings
  • Custom parameters
Most catalog MCPs are pre-configured and need minimal setup.
4

Add Justification

Explain why you need this MCP and what you’ll use it for
5

Submit Request

Click “Submit” - the request goes to your IT admin for approval
6

Wait for Approval

You’ll receive a notification when your admin reviews the request
Note: Catalog MCPs are typically approved faster as they’re pre-vetted.

​
Create Custom MCP

Best for: Internal MCPs not in the catalog.
1

Go to MCPs Page

Click “MCPs” in the sidebar, then “Add Server” → “Create New Server”
2

Fill Basic Information

  • Name: Descriptive name (e.g., “Company Knowledge Base MCP”)
  • Description: What this MCP does in 1-2 sentences
  • Category: Choose the appropriate category
3

Configure Connection

  • URL: The MCP server address (get this from the MCP developer)
    • Example: https://mcp.example.com
    • Or internal: http://internal-mcp.company.local:3000
  • Transport Type: Usually “Streamable HTTP” or “SSE”
4

Add Authentication (if needed)

  • Select auth type: OAuth, API Key, or None
  • Provide credentials if required
  • Configure OAuth settings if applicable
5

Set Environment Variables

Add any required configuration as key-value pairs
6

Submit for Review

Click “Create” to send to administrators
Note: Custom MCPs require thorough security review by your admin team.

​
After Submission

Approval Once submitted:
  1. Status shows “Pending” on the MCPs page
  2. Administrators are notified automatically
  3. You receive a notification with a request ID
  4. Wait for review - Check back for updates

​
Requesting Access to Existing MCPs

If an MCP exists but you don’t have access:
1

Verify It's Active

Check the MCP status is “Active” (not Pending or Inactive)
2

Contact Your Administrator

Send a message to your admin or use your organization’s request system
3

Provide Context

Explain:
  • Which MCP you need access to
  • Why you need it for your work
  • What tools you’ll use it with
4

Wait for Policy Update

Admin will create an access policy for you

​
Viewing Audit Logs

Audit Logs Every action in Runlayer is logged for security and compliance:
  1. Navigate to Audit Logs in the sidebar
  2. Filter by:
    • Date range
    • Action type (created, deleted, accessed, etc.)
    • User
    • Resource (specific MCP server)
  3. Click on any log entry to see full details

​
Additional Resources

Admin Handbook

If you’re an admin, see the complete admin guide

Security Best Practices

Comprehensive security guidelines for MCP usage

Adding Custom MCPs

Technical guide for creating and deploying custom MCP servers

Quick Start

Deploy and configure Runlayer for your organization

​
Common Issues & Questions

Quick fixes:
  1. Check MCP status is “Active” in Runlayer
  2. Reconnect the MCP (re-authenticate with OAuth if prompted)
  3. Restart your AI tool completely (quit and reopen)
  4. Contact admin if issue persists or you lack access
Quick fixes:
  1. Verify you clicked the connection button in Runlayer
  2. Restart your AI tool completely (quit and reopen)
  3. Check the MCP is “Active” status in Runlayer
  4. Try disconnecting and reconnecting
Quick fixes:
  1. Check Audit Logs for specific error details
  2. Verify you have permission for that tool
  3. Try reconnecting the MCP
  4. Contact admin if it persists - may need policy update
Use the “Create Custom Server” option to request it. Provide as much detail as possible:
  • Link to MCP documentation
  • What business problem it solves
  • What data it needs to access
  • Why existing MCPs don’t work
Some third-party services require domain whitelisting before they’ll accept connections. If you’re getting connection errors with these MCPs, your organization needs to whitelist Runlayer’s OAuth proxy.OAuth Proxy ConfigurationServices that commonly need whitelisting:
  • Atlassian (Jira, Confluence)
  • Datadog
  • Asana
  • HubSpot
  • Salesforce
Solution: Contact your admin and ask them to configure the OAuth proxy for these services. This allows Runlayer to act as a trusted intermediary, eliminating the need for individual domain whitelisting.Your admin will need to:
  1. Enable OAuth proxy in Settings
  2. Configure it for the specific service
  3. Add the proxy domain to the service’s allowlist
If you’re getting 403 errors when making legitimate requests, Runlayer’s security scanners may be flagging your activity as suspicious (false positive).What you’ll see:
Error: Request blocked - Security violation detected (403)
This can happen when:
  • Your request contains keywords that look like SQL injection
  • You’re querying sensitive data fields in bulk
  • Request patterns look like automated scanning
  • Your normal workflow triggers anomaly detection
What to do:
  1. Check audit logs - See exactly what was blocked and why
  2. Rephrase your request - Try asking in a different way
  3. Contact your admin - They can:
    • Review the false positive
    • Adjust scanner sensitivity
    • Whitelist your specific use case
    • Temporarily bypass scanners if needed
Note: Your admin can see the full context of blocked requests in the security alerts dashboard, making it easier to diagnose false positives.

​
Getting Help

  • Audit Logs: Check logs for error details and activity history
  • Documentation: Browse the full docs at your Runlayer instance
  • Email/Slack us: [email protected]
  • Admin Contact: Reach out to your Runlayer admin for MCP-specific questions