Unsplash API - FastAPI + FastMCP
Forked from unsplash-api by @aliosmankaya
Table of Contents
- Overview
- Prerequisites
- Installation
- Configuration
- Running
- API Endpoints
- MCP Integration
- Development
- License
Overview
This project provides an API to access the Unsplash service, allowing you to search, list, and get random images. Additionally, it integrates the Model Context Protocol (MCP), enabling AI models like Claude to interact directly with the Unsplash API.
Prerequisites
Before using the Unsplash API, you need to:
- Register as a developer on Unsplash
- Obtain your Access Key
- Configure the key as
UNSPLASH_CLIENT_IDin the.envfile
Installation
Using pip
# Clone the repository
git clone https://github.com/your-username/unsplash-api-mcp.git
cd unsplash-api-mcp
# Install dependencies
pip install -r requirements.txt
# Configure environment variables
cp .env.example .env
# Edit the .env file and add your UNSPLASH_CLIENT_ID
Using Docker
# Clone the repository
git clone https://github.com/your-username/unsplash-api-mcp.git
cd unsplash-api-mcp
# Configure environment variables
cp .env.example .env
# Edit the .env file and add your UNSPLASH_CLIENT_ID
# Build and start the container
docker compose up -d
Configuration
Create a .env file in the project root with the following content:
UNSPLASH_CLIENT_ID=your_access_key_here
Running
Locally
python main.py
The API will be available at http://localhost:8000.
With Docker
docker compose up -d
The API will be available at http://localhost:8000.
Access the interactive API documentation at http://localhost:8000/docs.
API Endpoints
Search
Endpoint to search for images on Unsplash.
Endpoint: /search
Method: GET
Parameters:
query: Search term (Default: "nature")page: Page number (Default: 1)per_page: Number of photos per page (Default: 10)order_by: Photo ordering (Default: "relevant", Options: "relevant", "latest")
Request Example:
GET /search?query=mountains&page=1&per_page=5&order_by=latest
Response Example:
[
{
"alt_description": "mountain range under cloudy sky",
"created_at": "2023-05-15T12:34:56Z",
"username": "Photographer Name",
"image_link": "https://images.unsplash.com/photo-...",
"download_link": "https://unsplash.com/photos/...",
"likes": 123
}
...
]
Photos
Endpoint to list photos from the Unsplash landing page.
Endpoint: /photos
Method: GET
Parameters:
page: Page number (Default: 1)per_page: Number of photos per page (Default: 10)order_by: Photo ordering (Default: "latest", Options: "latest", "oldest", "popular")
Request Example:
GET /photos?page=1&per_page=5&order_by=popular
Response Example:
[
{
"alt_description": "scenic view of mountains during daytime",
"created_at": "2023-06-20T10:15:30Z",
"username": "Photographer Name",
"image_link": "https://images.unsplash.com/photo-...",
"download_link": "https://unsplash.com/photos/...",
"likes": 456
},
...
]
Random
Endpoint to get random photos from Unsplash.
Endpoint: /random
Method: GET
Parameters:
query: Search term to filter random photos (Default: "nature")count: Number of photos to return (Default: 1, Maximum: 30)
Request Example:
GET /random?query=ocean&count=3
Response Example:
[
{
"alt_description": "blue ocean waves crashing on shore",
"created_at": "2023-04-10T08:45:22Z",
"username": "Photographer Name",
"image_link": "https://images.unsplash.com/photo-...",
"download_link": "https://unsplash.com/photos/...",
"likes": 789
},
...
]
For more information about the Unsplash API, see the official documentation.
MCP Integration
MCP Overview
The Model Context Protocol (MCP) is a protocol that allows AI models to interact directly with APIs and services. This implementation uses FastAPI-MCP to expose the Unsplash API endpoints as MCP tools.
MCP Endpoints
The MCP server is available at /mcp and exposes all API endpoints as MCP tools:
- search: Search for images on Unsplash
- photos: List photos from the landing page
- random: Get random photos
Using with AI Models
AI models that support MCP can connect to this API using:
http://your-server:8000/mcp
For Claude, you can configure the connection in the model settings or via API.
Example Client
You can test the MCP server with a simple Python client:
import requests
def test_mcp_metadata():
"""Test if the MCP server is working correctly."""
response = requests.get("http://localhost:8000/mcp/.well-known/mcp-metadata")
if response.status_code == 200:
print("MCP server working correctly!")
print(f"Response: {response.json()}")
else:
print(f"Error: {response.text}")
def list_mcp_tools():
"""List the available tools in the MCP server."""
response = requests.post(
"http://localhost:8000/mcp/jsonrpc",
json={
"jsonrpc": "2.0",
"id": 1,
"method": "mcp/list_tools"
}
)
if response.status_code == 200:
print("Available MCP tools:")
for tool in response.json()["result"]["tools"]:
print(f"- {tool['name']}: {tool['description']}")
else:
print(f"Error: {response.text}")
if __name__ == "__main__":
test_mcp_metadata()
list_mcp_tools()
For more information about using MCP, see the MCP_USAGE.md file.
Development
To contribute to development:
- Clone the repository
- Install development dependencies:
pip install -r requirements.txt - Create a
.envfile with your Unsplash API key - Run the server in development mode:
python main.py
License
This project is licensed under the MIT License - see the LICENSE file for details.