The mcp-searxng server is an MCP server that empowers AI models with web search capabilities via the SearxNG API. It allows models to securely access and leverage SearxNG's search functionalities, including general web queries, news, and articles, enhanced with pagination control for managing result size and offsets. The core tool, searxng_web_search, accepts search terms, results per page, and pagination offset as inputs. This server is implemented using JavaScript and Dockerfile, and integrates seamlessly with clients like Claude Desktop through Smithery or Docker. By using mcp-searxng, developers can equip their AI models with up-to-date information retrieval, enabling more informed and context-aware responses. Setting the SEARXNG_URL environment variable allows connection to public SearxNG instances or a local environment.

Web Search via SearXNG API

The mcp-searxng server provides a crucial web search capability to AI models by leveraging the SearXNG API. This allows models to access and process real-time information from the internet, enhancing their ability to answer questions, conduct research, and generate content based on current data. The server acts as an intermediary, securely relaying search queries from the AI model to a SearXNG instance and returning the results in a structured format that the model can readily understand. This integration is vital for AI applications that require up-to-date information, such as news summarization, market analysis, and question answering systems.

For example, an AI model tasked with writing a report on the latest advancements in renewable energy can use this feature to gather information from various online sources, ensuring the report is accurate and reflects the most recent developments. The mcp-searxng server abstracts away the complexities of interacting directly with web search engines, providing a clean and consistent interface for the AI model.

Technically, this is achieved through the searxng_web_search tool, which accepts a search query as input and returns a list of search results. The server is built using JavaScript and can be easily deployed using Docker, making it a portable and scalable solution for integrating web search into AI workflows.

Controlled Search Result Pagination

The pagination feature of mcp-searxng allows AI models and developers to precisely control the number of search results returned per request and the starting point of the results. This is essential for managing the volume of data processed by the AI model and optimizing performance. By specifying the count and offset parameters, users can retrieve search results in manageable chunks, avoiding overwhelming the model with excessive information. This fine-grained control is particularly useful when dealing with large datasets or when the AI model needs to iteratively refine its search strategy.

Consider a scenario where an AI model is tasked with identifying all research papers published on a specific topic in the last year. The model can use the pagination feature to retrieve the search results in batches of 20, starting from the most recent publications. This allows the model to efficiently process the information and avoid hitting rate limits imposed by the SearXNG instance.

The pagination is implemented within the searxng_web_search tool, allowing users to specify the count (results per page) and offset (pagination offset) parameters. The default values are set to 20 and 0, respectively, but can be adjusted to suit the specific needs of the AI model.

Secure and Standardized MCP Interface

mcp-searxng provides a secure and standardized interface for AI models to access web search capabilities, adhering to the MCP (Model Context Protocol) standards. This ensures that the interaction between the AI model and the SearXNG search engine is governed by a well-defined protocol, promoting interoperability and security. By using MCP, the server can securely authenticate requests from the AI model, prevent unauthorized access to the search engine, and ensure that the data exchanged between the two systems is protected. This is particularly important when dealing with sensitive information or when the AI model is deployed in a production environment.

For instance, an AI-powered customer service chatbot can use the mcp-searxng server to quickly find answers to customer queries by searching the web. The MCP interface ensures that the chatbot can securely access the search engine without exposing sensitive customer data or compromising the security of the system.

The server leverages standard input/output streams or HTTP/SSE for communication, making it compatible with a wide range of AI models and platforms. The use of MCP also simplifies the integration process, as developers can rely on a consistent and well-documented protocol for interacting with the search engine.

Integration Flexibility

mcp-searxng offers flexible integration options, supporting both direct installation via Smithery and containerization via Docker. This allows developers to choose the deployment method that best suits their environment and workflow. The Smithery installation provides a streamlined process for automatically setting up the server for use with Claude Desktop, while the Docker option offers greater portability and scalability. This dual approach ensures that mcp-searxng can be easily integrated into a variety of AI development environments, regardless of the underlying infrastructure.

For example, a developer working on a local machine can use the Smithery installation to quickly set up the server and start experimenting with web search integration. Alternatively, a team deploying an AI model to a cloud platform can use the Docker image to ensure consistent and reproducible deployments across different environments.

The Dockerfile provided in the repository simplifies the process of building and deploying the server, while the NPX configuration allows for seamless integration with Claude Desktop. This flexibility makes mcp-searxng a versatile solution for integrating web search into AI applications.