MCP Servers for Claude & ChatGPT
PortEden provides MCP (Model Context Protocol) servers that give Claude, Claude Cowork, and ChatGPT secure access to your email, calendar, drive, and task management tools, with the same granular rules you'd set for any other connection type.
What Are MCP Servers?
MCP (Model Context Protocol) is a standard that lets AI platforms connect to external data sources and tools. Instead of giving AI raw OAuth tokens, an MCP server provides a structured, permission-controlled interface. PortEden's MCP servers act as data firewalls. Every request from AI passes through your access rules before any data is returned.
Note
Supported AI Platforms
PortEden MCP servers work with any platform that supports the Model Context Protocol:
Claude
Connect Claude Desktop or Claude API to PortEden via MCP for secure email, calendar, drive, and task access.
Claude Cowork
Claude Cowork connects through MCP servers, giving your team AI-powered workflows with full security controls.
ChatGPT
ChatGPT supports MCP connections, allowing PortEden to filter every data request through your access rules.
Available Services via MCP
PortEden MCP servers expose the same services available through the REST API and OpenClaw skills:
- Email: Search, read, send, reply, and forward across Gmail, Outlook, and Exchange
- Calendar: List, create, update events across Google Calendar and Outlook Calendar
- Drive: Search, read, upload files across Google Drive, Docs, and Sheets
- Task Management: List, create, update tasks across Jira, Confluence, Monday.com, Asana, Linear, and Notion
Note
How It Works
Create an Access Token
Sign in at my.porteden.com or use the Authentication API to generate an access token. Configure permissions, access rules, and connected services.
Configure Your MCP Client
Point your AI platform's MCP configuration to PortEden's MCP server endpoint. Provide your access token as the authentication credential. The AI platform will discover available tools automatically.
AI Accesses Data Through Your Rules
When Claude or ChatGPT makes a request through the MCP server, PortEden enforces your access rules: visibility controls, contact rules, action limits, time windows, and board restrictions. Blocked requests are logged. Allowed data is returned in clean, minimal format.
MCP Servers vs OpenClaw Skills
Both connection methods provide the same security features. Choose based on your AI platform:
| Feature | MCP Servers | OpenClaw Skills |
|---|---|---|
| Platforms | Claude, Claude Cowork, ChatGPT | OpenClaw-compatible agents, Copilot, Gemini |
| Security Rules | Same permissions, access rules, and audit trails | Same permissions, access rules, and audit trails |
| Setup | Configure MCP endpoint in AI platform settings | Install OpenClaw skill via CLI |
| Services | Email, Calendar, Drive, Tasks | Email, Calendar, Drive, Tasks |
| Context Hygiene | Yes, clean, minimal responses | Yes, clean, minimal responses |
Security
MCP servers enforce the same six layers of access control as every other PortEden connection:
- Visibility: Control how much data AI can see (free/busy only, headers only, full content)
- Contact Rules: Block data from specific people or domains
- Action Limits: Restrict to read-only, draft-only, or specific operations
- Time Windows: Limit access to recent data only
- Account & Board Scope: Restrict to specific accounts, boards, or workspaces
- Data Reduction: Redact sensitive fields from responses
Every request through MCP is logged in the audit trail. You can revoke access with one click at any time.