A new open-source tool brings a 13-agent research team and 10-stage pipeline to Claude Code users tackling academic papers. Academic Research Skills (ARS) handles reference hunting, citation formatting, data verification, and logical consistency checks — but explicitly refuses to write the paper for you. The tool installs via plugin marketplace and costs roughly $4-6 for a full 15k-word paper.

The architecture separates research (13 agents), writing (12 agents with Style Calibration and Writing Quality Check), and peer review (7 agents with quality rubrics). Stage 2.5 and Stage 4.5 integrity gates run a 7-mode blocking checklist designed to catch implementation bugs, hallucinated results, and citation fabrications. The system draws on Lu et al.'s 2026 Nature paper documenting failure modes in fully autonomous AI research systems.

The project builds on a clear premise: human researchers augmented by AI avoid the failure modes that tripped The AI Scientist (which scored 6.33/10 at ICLR 2025). Rather than replacing the researcher, ARS aims to handle the tedious groundwork so academics can concentrate on defining research questions, selecting methodologies, and crafting arguments — the intellectual work that demands human judgment.