Two recent systems treat published papers as machine-actionable substrates rather than static text: Huang et al. 2025's Biomni mines tens of thousands of papers to extract tasks, tools and databases into a unified action space, while Miao et al. 2026's Paper2Agent converts individual papers into MCP-based agents validated to reproduce the original results and answer new queries.
E9 describes constructing a unified action space by analyzing tens of thousands of papers to extract tasks, tools, and databases, and E34 describes Paper2Agent converting papers into MCP-server-based agents validated to reproduce original results and carry out novel user queries.
Written by Kimi K3 via Ollama Cloud · checked by GLM-5.3 via Ollama Cloud · 30 Sept, 23:08
Source chain
AEvery quote below was checked, without a model, to appear verbatim in its source.
- 01
“agent capable of reading papers and extracting key tasks, tools, and databases”
Full-text passage · p. 4 · evidence E9
action space, which is inherently diverse, domain-specific, and complex. Moreover, a truly capable54 system requires an agentic architecture that can natively interact with this biomedical environment55 – autonomously selecting and composing actions, using its reasoning capabilities to plan and exe-56 cute diverse tasks without relying on rigid, pre-defined workflows.57 Here we present Biomni, a generalist biomedical AI agent purpose-built to automate and58 advance biomedical research across a wide range of subfields. Acting as a virtual AI biologist,59 Biomni autonomously formulates novel, testable hypotheses, performs complex bioinformatics60 analyses, and designs rigorous experimental protocols. To enable this capability, we first con-61 structed a unified and comprehensive biomedical action space by systematically analyzing tens62 of thousands of biomedical research papers spanning 25 distinct subfields, curated from major63 bio-literature repositories. From this foundation, we developed an LLM-powered action discovery64 agent capable of reading papers and extracting key tasks, tools, and databases essential to driving65 biomedical discoveries.
Passage read from www.biorxiv.org, which may be a preprint rather than the published version.
- 02
“We validate that these agents reproduce the results of the original papers and carry out novel user queries”
Abstract · no page number · evidence E34