MCP Overview

The Tampa.dev Platform provides a Model Context Protocol (MCP) server, enabling AI assistants, personal automation tools, and agent frameworks to interact with the Tampa.dev community programmatically.

What is MCP?

The Model Context Protocol is an open standard for connecting AI assistants to external tools and data sources. It defines a JSON-RPC 2.0 interface that allows AI models to discover and invoke tools, read structured resources, and use prompt templates -- all through a single, well-defined protocol.

MCP eliminates the need for AI agents to understand REST API conventions, pagination, or authentication flows in detail. Instead, agents discover available capabilities at runtime and invoke them through a uniform interface.

Why Tampa.dev supports MCP

The Tampa.dev MCP server allows AI agents to:

Browse events and groups -- search upcoming events, view group details, check leaderboards
Manage RSVPs and check-ins -- RSVP to events, check in with codes, view attendee lists
Interact with profiles -- read and update profile information, view achievements and badges
Manage favorites and follows -- add/remove favorite groups, follow/unfollow users
Administer groups -- create events, manage members, award badges, generate check-in codes
Run platform admin tasks -- manage users, sync data, configure achievements (admin only)

All of this is governed by the same scope system and authentication used by the REST API, so agents operate with the same permissions as the user who authorized them.

Server Details

Property	Value
Server URL	`https://api.tampa.dev/mcp`
Transport	Streamable HTTP (`POST /mcp` with JSON-RPC 2.0)
Protocol version	`2025-03-26`
Discovery	`https://api.tampa.dev/.well-known/mcp-configuration`

Quick Start

1. Discover the server

Fetch the well-known MCP configuration to confirm the server endpoint and capabilities:

curl https://api.tampa.dev/.well-known/mcp-configuration

{
  "mcpServers": {
    "tampa-devs": {
      "url": "https://api.tampa.dev/mcp",
      "transport": "streamable-http",
      "authentication": {
        "type": "oauth2",
        "authorizationUrl": "https://tampa.dev/oauth/authorize",
        "tokenUrl": "https://tampa.dev/oauth/token",
        "scopes": ["read:events", "read:groups", "read:user"]
      }
    }
  }
}

2. Authenticate

MCP uses the same authentication as the REST API:

OAuth 2.1 with PKCE -- the recommended flow for interactive AI clients. See Authentication.
Personal Access Token -- simpler option for scripts and local tools. See Personal Access Tokens.

Include the token in the Authorization header of every POST /mcp request:

Authorization: Bearer td_pat_abc123...

3. Initialize the session

Send an initialize request to establish the MCP session:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "initialize",
  "params": {
    "protocolVersion": "2025-03-26",
    "capabilities": {},
    "clientInfo": {
      "name": "my-ai-agent",
      "version": "1.0.0"
    }
  }
}

The server responds with its capabilities:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "protocolVersion": "2025-03-26",
    "capabilities": {
      "tools": {},
      "resources": {},
      "prompts": {}
    },
    "serverInfo": {
      "name": "tampa-devs-mcp",
      "version": "1.0.0"
    }
  }
}

4. Discover tools

List the tools available to your token's scopes:

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/list"
}

The response includes only the tools your token's scopes allow. See Scope-filtered discovery below.

5. Call a tool

Invoke a tool by name with the required arguments:

{
  "jsonrpc": "2.0",
  "id": 3,
  "method": "tools/call",
  "params": {
    "name": "events_list",
    "arguments": {
      "status": "upcoming",
      "limit": 5
    }
  }
}

Capabilities

The Tampa.dev MCP server exposes three capability types:

Capability	Description	Documentation
Tools	Executable actions (read data, create resources, manage settings)	MCP Tools
Resources	Read-only data accessible via `tampadev://` URIs	MCP Resources
Prompts	Reusable prompt templates for common workflows	MCP Prompts

Scope-Filtered Discovery

When an AI agent calls tools/list, the MCP server inspects the caller's token scopes and returns only the tools the caller is authorized to use. This means:

A token with read:events will see event browsing tools but not event management tools.
A token with manage:groups and manage:events will additionally see group and event management tools.
A token with admin will see the full platform administration tool set.
Session-authenticated users (first-party web UI) see all tools, since session auth bypasses scope restrictions.

This design prevents AI agents from attempting operations they cannot perform, reducing unnecessary error handling and improving agent reliability.

Authentication and Scopes

MCP uses the same scope system as the REST API. Request only the scopes your agent needs:

Agent Use Case	Recommended Scopes
Event browser	`read:events`, `read:groups`
Personal assistant	`read:events`, `read:user`, `read:favorites`, `write:favorites`
RSVP manager	`read:events`, `write:events`
Group organizer	`manage:groups`, `manage:events`, `manage:checkins`, `manage:badges`
Full admin bot	`admin`

For the complete scope reference, see Scopes.

Next Steps

MCP Tools -- full tool reference organized by domain
MCP Resources -- read-only data resources for AI context
MCP Prompts -- reusable prompt templates
MCP Examples -- end-to-end integration examples

API Documentation