What is multi-search-aggregator?

This is a tool that runs a single search query across more than 30 different search engines at the same time, then merges the results into one ranked list. Instead of picking one search source and hoping it covers the topic, it fires all of them in parallel and combines what comes back. The README says it was designed specifically for AI agents and research workflows where broad coverage matters. The engines span a wide range. General web search includes Google, Bing, Brave, Startpage, and Qwant. Academic sources include arXiv, OpenAlex, and DBLP. Developer-focused sources include Stack Overflow, npm, and MDN. Social platforms include Reddit and Hacker News. There are also several Chinese-language engines and AI-powered search services. Any engine that lacks an API key is simply skipped rather than causing an error. Results from all engines are deduplicated by URL, then scored using a combination of keyword relevance, how authoritative a given source is considered to be, how recent the result is, and vector-based semantic similarity. The top results are returned in a single list. A web interface (on port 8200 by default) lets you search through a browser with an engine filter sidebar. A REST API provides the same results as JSON. A streaming option sends progress updates in real time as each engine finishes, so you can see results arriving rather than waiting for the full batch. Configuration lives in a YAML file where you set engine weights, timeouts, and which engines are enabled. Changes take effect without restarting the server. Adding a new engine means writing a Python function that takes a query string and returns a list of result dictionaries, then registering it in a single file. Setup requires Python 3.10 or later. You clone the repo, install the Python dependencies, copy the example environment file, and add API keys for whichever engines you want. A command-line interface is also available for one-off searches without starting the web server.