Copy page

AI & MCP · New

MCP Integration

The IntelCenter API supports the Model Context Protocol (MCP), enabling AI systems such as ChatGPT, agents, and automation tools to directly access and interact with IntelCenter intelligence data.

MCP allows structured tool-based access to the API without manually constructing HTTP requests.

---

What is MCP?

MCP (Model Context Protocol) enables AI systems to:

• Discover available tools dynamically
• Call API-backed functions
• Pass structured parameters
• Receive structured responses

With MCP, IntelCenter data can be queried and analyzed naturally within AI workflows.

---

Available MCP Endpoints

ICD Only

POST /mcp-icd

Provides access to ICD datasets:

• ICD Video
• ICD Image
• ICD Natural Language Search
• ICD Domain GenAI

---

Full Dataset

POST /mcp-all

Provides access to:

• ICD Video
• ICDB Video
• ICD Image
• ICDB Image
• ICD Natural Language Search
• ICDB Natural Language Search
• ICD Domain GenAI
• ICDB Domain GenAI

---

Available Tools

Search Tools

• searchIcdVideos
• searchIcdbVideos
• searchIcdImages
• searchIcdbImages

Natural Language Search

• searchIcdNl
• searchIcdbNl

Domain GenAI

• icdDomainGenAiChat
• icdbDomainGenAiChat

Record Retrieval

• getOneIcdVideo
• getOneIcdbVideo
• getOneIcdImage
• getOneIcdbImage

Media Access

• getIcdVideoPlayer
• getIcdbVideoPlayer
• getIcdImageViewer
• getIcdbImageViewer

---

Supported Parameters

Standard Search Parameters

Search tools accept the same parameters as REST endpoints:

Parameter	Description
search	Full text search
entity	Filter by entity
country	Filter by country
page	Page number
per_page	Results per page (capped by your API key’s limit)
date_from	Start date (YYYY-MM-DD)
date_to	End date (YYYY-MM-DD)
sort	Optional sorting

---

Natural Language Search Parameters

Parameter	Description
search	Natural language query
entity	Optional entity filter
country	Optional country filter
source_component	Filter by content type
top_k	Retrieval depth
page	Page number
per_page	Results per page

---

Domain GenAI Parameters

Parameter	Description
question	Required question to answer
entity	Optional entity focus
primary_country	Optional country focus
conversation_id	Optional conversation context
top_k	Retrieval depth

---

Example MCP Usage

In an MCP-enabled environment (such as ChatGPT), a tool call may look like:

Code
{
  "tool": "searchIcdbVideos",
  "arguments": {
    "search": "houthis",
    "page": 2,
    "per_page": 5
  }
}

Example Domain GenAI call:

Code
{
  "tool": "icdDomainGenAiChat",
  "arguments": {
    "question": "What recent naval drone activity has been observed?",
    "top_k": 5
  }
}

The system will automatically call the IntelCenter API and return structured results.

---

Example Response (Simplified)

Code
{
  "items": [
    {
      "uuid": "11111111-1111-1111-1111-111111111111",
      "title": "Example Video Title",
      "primary_date": "2026-03-12",
      "entity": "Example Entity"
    }
  ],
  "curPage": 2,
  "perPage": 5
}

---

Mapping to REST API

Each MCP tool maps directly to a REST endpoint:

MCP Tool	REST Endpoint
searchIcdbVideos	/v1/icdb/videos
searchIcdVideos	/v1/icd/videos
searchIcdbImages	/v1/icdb/images
searchIcdImages	/v1/icd/images
searchIcdbNl	/v1/icdb/nl-search
searchIcdNl	/v1/icd/nl-search
icdbDomainGenAiChat	/v1/icdb/domain-genai
icdDomainGenAiChat	/v1/icd/domain-genai
getOneIcdbVideo	/v1/icdb/videos/{uuid}
getOneIcdVideo	/v1/icd/videos/{uuid}
getIcdbVideoPlayer	/v1/icdb/videos/{uuid}/player
getIcdVideoPlayer	/v1/icd/videos/{uuid}/player
getIcdbImageViewer	/v1/icdb/images/{uuid}/viewer
getIcdImageViewer	/v1/icd/images/{uuid}/viewer

---

Authentication

MCP endpoints use the same authentication as REST:

Code
Authorization: Bearer YOUR_API_KEY

---

Access Control

MCP tools are dynamically available based on your API key permissions.

This means:

• Some tools may not be accessible depending on your dataset access
• ICDB tools may be restricted to specific entities
• Natural Language and Domain GenAI tools may not be enabled for all keys

If a tool is not permitted, requests will return a 403 Forbidden response.

---

Usage Limits

MCP usage is subject to the same limits as standard API requests:

• Rate limits (search and detail endpoints)
• Domain GenAI limits (per-minute, hourly, daily)
• Daily quotas
• Monthly quotas
• Per-request limits (per_page)

Each MCP tool invocation counts as an API request.

If limits are exceeded, the API will return a 429 Too Many Requests response.

---

When to Use MCP

Use MCP when:

• Integrating IntelCenter into AI assistants (ChatGPT, agents)
• Building automated workflows
• Enabling natural language querying and analysis

Use REST when:

• Building traditional applications
• Working with direct HTTP requests
• Needing full control over request/response handling

---

Notes

• MCP automatically handles parameter mapping and request formatting
• Responses follow the same structure as REST endpoints
• Pagination, filtering, and sorting behave identically
• Designed for structured, reliable AI integration
• Tool availability is permission-aware

---

Getting Started

To begin using MCP:

1. Connect to the /mcp-all endpoint
2. Authenticate using your API key
3. Discover available tools
4. Call tools with structured parameters
5. Process structured responses in your application or AI workflow

---

Support

For assistance integrating MCP:

[email protected]