Paper Lantern

Paper Lantern is an open-source academic-search tool that gives coding agents access to the latest research techniques instead of relying only on their training data. It searches a corpus of more than two million computer science papers and surfaces the methods most relevant to the task an agent is working on — for example, the right approach for writing stronger tests or extracting clauses from a contract. The goal is measurable: on a mutation-score testing task, baseline accuracy of 63% rose to 87% when the agent was guided by Paper Lantern, and on a contract-analysis task, accuracy improved from 44% to 76%. Because it is open-source, teams can inspect how it works and run it themselves rather than depending on a closed service. Paper Lantern is aimed at developers and AI engineers building agentic software-engineering workflows who want their agents to reason with current research rather than reinventing known techniques. Coverage is centered on computer science, so its strongest fit is technical tasks where the academic literature offers concrete, applicable methods.

Developers and AI engineers building coding agents that need to apply proven research techniques to tasks like test generation and clause extraction.