What is read-summarize-papers?

read-summarize-papers, called rsp for short, is a command line tool that helps researchers keep up with new academic papers without manually copying anything. It connects to your Zotero library, can automatically pull in new papers from arXiv based on keyword searches you define, and then uses an AI language model to read and analyze each paper against a set of questions you write in advance about your own research interests. The tool runs a two tier process to save cost and time. Papers are first screened against your research profile using a simple keyword match. Papers that only weakly match get a cheaper summary based on just the abstract, while papers that strongly match get a full analysis that reads the whole paper, answers your custom questions with evidence, and adds the paper to a visual research map. Every result is written as a structured note in Obsidian, organized by the date it was processed, so your notes build into a browsable archive over time. Beyond the main pipeline, rsp includes commands to find and merge duplicate entries in your library, attach PDF copies to your Zotero items, sort papers into collections based on keyword matches, search through everything you have processed, and check that your setup is configured correctly. Interrupted runs are safe to resume, since already processed papers are skipped and only the interrupted one is retried. To use it you need Python 3.9 or later, a Zotero account with an API key, an Obsidian vault, and an API key for an AI service such as Anthropic or any OpenAI compatible provider. Setup involves copying example configuration files, filling in your paths and API keys, and defining the arXiv sources and questions you care about. Once configured, a typical day involves running one command to fetch or sync new papers and another to process everything that is pending.