Academic Research Skills
github.com/imbad0202/academic-research-skills- Category
- AI Agents
- Rank
- No. 55Tools index
- Listed in
- #2 Research with an agent
- Pricing
- Open Source
- Platform
- cli
- Type
- AGENT
- Builder
- imbad0202
- GitHub
- 47.1k stars
- Latest release
- v3.21.2
- Date
About
A comprehensive suite of Claude Code skills for academic research, covering the full pipeline from literature review to publication. Features AI-powered research agents, peer review automation, citation verification, and integrity checking to help researchers focus on insights while handling the grunt work.
What it does
Academic Research Skills organizes a paper project into staged, user-approved work. It can guide question formation, search and verify references, structure drafts, coordinate specialist reviewers, revise against feedback, and prepare final formats. A Material Passport carries declared evidence and checkpoint state between stages or sessions.
Why it's ranked here
The strongest case is its explicit control structure. Every stage requires confirmation, integrity gates can block progress, reviewers work under defined constraints, and citation existence checks query four bibliographic indexes. The caution is equally concrete: corpus-scale evaluation remains future work, while a full run can consume substantial tokens and span hours or days.
What's good
It treats provenance and research judgment as first-class workflow concerns. Citation checks use external bibliographic services and a local cache. Review includes paper-blind scoring commitments before reviewers see the manuscript. Existing literature collections can enter through a declared passport, then face the same inclusion rules and verification path as newly found sources.
Tradeoffs
The full workflow is expensive in context and attention, with mandatory checkpoints and potentially long review loops. Core operation depends on Claude Code and an Anthropic API key. Direct document and PDF output require extra tools. Journal-tier judgments remain advisory model knowledge, and the supplied corpus adapters neither extract PDF text nor contact remote library services.
How to use it well
It suits researchers who want structured assistance while retaining responsibility for questions, methods, evidence, and final claims. Start with one focused mode, then adopt the full pipeline when its checkpoints justify the cost. Bring curated references through a Material Passport. Use separate tooling for OCR, live Zotero synchronization, institutional-library access, or persistent knowledge across papers.
Technical notes+
README.md describes four prompt-driven skills, slash-command entry points, integrity gates, and plugin installation. docs/ARCHITECTURE.md defines the staged pipeline, human checkpoints, data-access levels, agent roles, and artifact flow. docs/SETUP.md documents optional Python adapters, Pandoc and tectonic output dependencies, cross-model flags, and a SQLite citation-verification cache. docs/PERFORMANCE.md records high token budgets, inherited model routing, restricted plugin-agent tool allowlists, and passport-based session resume. package.json marks the package private, applies CC-BY-NC-4.0, and declares Pi skills and prompts. pyproject.toml only supplies pytest configuration.
Observed
- License
- CC-BY-NC-4.0
- Packaging
- Private package with Claude Code plugin installation and a Pi package surface
- Interfaces
- Claude Code skills and slash commands; Pi package support is also declared
- Core composition
- Four prompt-driven skills cover deep research, paper writing, paper review, and pipeline orchestration
- Platform support
- Claude Code CLI, VS Code, and JetBrains receive full plugin support; Claude Science imports only the skill methodology
- Optional dependencies
- Pandoc enables direct DOCX output, tectonic enables PDF output, and Python supports selected adapters and utilities
Read from README.md, package.json, pyproject.toml, docs/SETUP.md, docs/PERFORMANCE.md, docs/SETUP.zh-TW.md, docs/ARCHITECTURE.md, docs/ROADMAP-v3.11.md, docs/PERFORMANCE.zh-TW.md, docs/cross-paper-workflow.md, docs/design/2026-05-05-phase-6.6-scoping-note.md.
What it can do
Conduct automated literature review
Research topic or keywords → Comprehensive literature summary with relevant papers
Verify citations and references
Academic paper with citations → Citation accuracy report with corrections
Generate academic papers
Research data and topic requirements → Structured academic paper draft
Perform automated peer review
Academic paper manuscript → Detailed peer review report with feedback
Check research integrity
Research manuscript or data → Integrity assessment report with recommendations
Orchestrate multi-stage research pipeline
Research project parameters → Coordinated research workflow execution results
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.