MCP (Model Context Protocol)
Manage Model Context Protocol servers to extend your assistants with custom functions and external integrations.What is MCP?: The Model Context Protocol allows your assistants to call external functions, query databases, integrate with APIs, and perform custom business logic. Learn more in our MCP Guide.
Overview
MCP endpoints allow you to:- Register external function servers
- Connect functions to specific assistants
- Manage function availability and permissions
- Monitor MCP server status and health
Authentication
All MCP endpoints require authentication via API key or Bearer token.Core Workflow
1
Register MCP Server
Create and register your MCP server with Plati
2
Connect to Assistant
Attach the MCP functions to specific assistants
3
Use in Prompts
Functions become available as
{{function.name()}} in prompts4
Monitor & Manage
Track function usage and manage connections
Endpoints
Create MCP Server
Register a new Model Context Protocol server in your workspace.List MCP Servers
Get all registered MCP servers in your workspace.Connect MCP to Assistant
Attach an MCP server to a specific assistant to make its functions available.Delete MCP Server
Remove an MCP server from your workspace.Request/Response Examples
MCP Registration Response
Connection Response
MCP Server Requirements
Your MCP server must implement these endpoints:GET /tools
Return available functions and their schemas.POST /call
Execute function calls from assistants.Error Handling
400 Bad Request
400 Bad Request
401 Unauthorized
401 Unauthorized
404 Not Found
404 Not Found
502 Bad Gateway
502 Bad Gateway
Best Practices
Security
- Always use HTTPS for MCP servers
- Implement proper authentication
- Validate all function inputs
- Rate limit function calls
Performance
- Keep function responses under 1MB
- Implement caching where appropriate
- Use async patterns for long operations
- Monitor function execution times
Reliability
- Handle errors gracefully
- Provide clear function descriptions
- Test all function combinations
- Implement health check endpoints
Monitoring
- Log all function calls
- Monitor success/failure rates
- Track function usage patterns
- Set up alerting for failures
