Academic Research Skills for Claude Code: A Human-in-the-Loop Pipeline for Paper Writing

Academic Research Skills (ARS) for Claude Code is a plugin that supports the full research-to-publication pipeline: research → write → review → revise → finalize. It is designed as a human-in-the-loop system, explicitly rejecting full automation. The tool handles grunt work — reference hunting, citation formatting, data verification, logical consistency checks — while the researcher retains control over question definition, method selection, interpretation, and the core argument.

Quick Install (Claude Code v3.7.0+)

/plugin marketplace add Imbad0202/academic-research-skills
/plugin install academic-research-skills

Verify with /ars-plan to start a Socratic dialogue mapping out chapter structure, or /ars-lit-review "your topic" for a single-shot literature review.

Why Human-in-the-Loop?

ARS cites Lu et al. (2026, Nature 651:914-919) who built The AI Scientist — the first fully autonomous AI research system to publish a paper through blind peer review at a top ML venue (ICLR 2025 workshop, score 6.33/10 vs workshop average 4.87). Their Limitations section lists failure modes: implementation bugs, hallucinated results, shortcut reliance, bug-as-insight reframing, methodology fabrication, frame-lock, and citation hallucinations. ARS is built on the premise that a human researcher augmented by AI avoids these failure modes better than either alone.

Integrity Gates & Calibration

Stage 2.5 and Stage 4.5 run a 7-mode blocking checklist (see academic-pipeline/references/ai_research_failure_modes.md). The reviewer offers an opt-in calibration mode that measures its own false negative rate / false positive rate against a user-supplied gold set.

Features

• Style Calibration learns your voice from past work.
• Writing Quality Check catches patterns that make prose feel machine-generated.
• Semantic Scholar API verification (inspired by PaperOrchestra, Song, Song, Pfister & Yoon, 2026, Google).
• Anti-leakage protocol, VLM figure verification, and score trajectory tracking.

Architecture

Full pipeline documentation is in docs/ARCHITECTURE.md, including flow diagram, stage-by-stage matrix, data-access flow, skill dependency graph, quality gates, and mode list. For DOCX output, pandoc is required; for APA 7.0 PDF, tectonic + Source Han Serif TC font (Markdown output works without either).