An AI-powered research assistant that performs iterative, deep research on any topic by combining search engines, web scraping, and large language models. It refines its research direction over time, dives deeper into topics, and produces comprehensive markdown reports with findings and sources.
Performs iterative, deep research on any topic by generating search queries, processing results, and diving deeper based on findings, then produces a comprehensive markdown report with findings and sources. Intended to be the simplest implementation of a deep research agent that can refine its research direction over time and deep dive into a topic.
The agent is an iterative deep-research assistant whose output is a comprehensive markdown report. Per the README and the code-derived surface analysis, it generates search queries, scrapes/reads web content via Firecrawl, performs LLM inference, and recursively refines research direction — but its terminal product is a document, not actions on external systems. No subprocess/shell, exec/eval, or autonomous action signals were detected in the entrypoint; outbound network calls are read-only (search/extraction + LLM inference). The autonomous recursive loop (depth/breadth) would otherwise suggest Tier 4, but because the agent is advisory by design (produces a report) it is capped at Tier 2. It does collect initial query/parameters and follow-up answers from the user, but this is input gathering, not a code-enforced approval gate before consequential actions.