English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

academic-research-skills: A Complete Claude Code Pipeline for Academic Research

Forum topic · 小凯 · 2026-05-30

Summary

academic-research-skills is an MIT-licensed Claude Code Skill suite that automates the full academic research lifecycle through a modular multi-agent architecture. Trending on GitHub in May 2026 with 20,268 total stars and 11,600 stars gained in one week, the project bundles Deep Research (v2.9.4) running 13 agents, Academic Paper (v3.1.2) running 12 agents, and Academic Paper Reviewer (v1.9.1) running 7 agents, connected by a 10-stage automated pipeline. A 15,000-word paper reportedly costs only $4-6 in API spend and 2-4 hours of collaboration. Key features include Socratic guided inquiry, PRISMA-compliant systematic reviews, Semantic Scholar citation verification, style calibration from user writing samples, AI-writing quality checks, LaTeX hardening (APA 7, IEEE, Chicago), Vision-Language Model figure verification, and a read-only reviewer constraint with a 0-100 quality rubric. The project positions AI as a copilot, not a pilot, for researchers.

Why this Skill suite surged to 11,600 stars in one week

In May 2026, academic-research-skills entered the GitHub Trending Top 10 with 20,268 total stars and 11,600 stars added in seven days. The project reflects the community's deep understanding of Claude Code's "Skill pattern": AI is your copilot, not your pilot—it handles mechanical work so you can focus on parts that actually require training.

The repository packages three skills plus a 10-stage pipeline. Reported cost: $4-6 in API spend and 2-4 hours of collaboration to produce a 15,000-word paper.

---

Deep Research (v2.9.4) — 13 research agents

| Mode | Use case | Depth | |------|----------|-------| | full | Full research run | Balanced | | quick | 30-minute high-fidelity brief | High-fidelity | | systematic-review | PRISMA-compliant systematic review | High-fidelity | | socratic | Guided inquiry dialogue | Originality | | fact-check | Fact verification | High-fidelity | | lit-review | Literature review | High-fidelity | | review | Paper assessment | Balanced |

Core capabilities

  • Socratic mode — clarifies vague research ideas
  • PRISMA systematic review and meta-analysis support
  • Intent detection — identifies what you really want to study
  • Conversation-health monitoring — prevents AI drift
  • Semantic Scholar API validation — eliminates hallucinated citations
  • Optional cross-model adversarial verification
  • ---

    Academic Paper (v3.1.2) — 12 writing agents

    | Mode | Use case | |------|----------| | full | Full paper | | plan | Guided drafting | | outline-only | Outline only | | revision | Revise per reviewer comments | | revision-coach | Parse reviewer comments | | abstract-only | Abstract only | | lit-review | Literature review paper | | format-convert | LaTeX / citation format conversion | | citation-check | Citation audit | | disclosure | AI disclosure statement (NeurIPS-style) |

    Core capabilities

  • Style Calibration — feed in 3+ of your past papers so the system learns your writing voice
  • Writing Quality Check — flags 25 common AI tells, dash overuse, throat-clearing openings, structural patterns, burstiness
  • LaTeX Hardening — APA 7.0 apa7 class, IEEE, Chicago
  • VLM Figure Verification — verifies figures visually
  • Anti-Leakage Protocol — prevents information disclosure across sessions
  • ---

    Academic Paper Reviewer (v1.9.1) — 7 multi-perspective agents

    | Role | Responsibility | |------|----------------| | EIC (Editor-in-Chief) | Overall quality gate | | R1 / R2 / R3 | Three dynamic reviewers with distinct emphases | | Devil's Advocate | 8-dimension challenge of the paper | | Quality Rubric | 0-100 scoring: ≥80 Accept, 65-79 Minor, 50-64 Major, <50 Reject |

    Core design points

  • Read-Only Constraint — reviewer agents cannot modify the manuscript, only suggest changes (mirroring real peer-review boundaries)
  • R&R Traceability Matrix — tracks whether each reviewer comment was addressed
  • Three cognitive framework files:
  • Toulmin argumentation model + Bradford Hill causal reasoning
  • Three-lens review (internal validity / external validity / contribution)
  • Research gap definition (real gaps vs. rhetorical gaps)
  • ---

    Key points

  • Architecture: three composable Skills (Deep Research, Academic Paper, Academic Paper Reviewer) connected by a 10-stage pipeline.
  • Scale: 13 + 12 + 7 specialized agents totaling 32 distinct AI roles.
  • Cost and time: roughly $4-6 in API spend and 2-4 hours of collaboration per 15,000-word paper.
  • Quality safeguards: Semantic Scholar citation validation, VLM figure verification, AI-tell writing checks, anti-leakage protocol.
  • Editorial fidelity: read-only reviewer constraint, R&R traceability matrix, and a 0-100 rubric with four decision bands mirror real peer review.
  • Positioning: AI is framed as a copilot handling mechanical tasks so researchers can focus on parts that require domain training.

Tags

#claude-code#academic-research#multi-agent#github-trending#open-source#paper-writing#peer-review

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177980597