Imbad0202 /
academic-research-skills
Claude Code skills that guide a paper from research through writing, review and revision
- 53.5k stars
- 3.4k stars since 3 Oct
- Python
- Updated 2 days ago
- Some experience
- Agent skills & plugins
Writing a paper involves many separate chores: finding sources, formatting citations, checking that claims match their sources, and answering reviewers. This project packages those steps as Claude Code skills you run with slash commands from the CLI or an IDE.
The pipeline runs from research to writing, review, revision and finalization, and it stops for your approval at each stage. Two integrity gates (Stages 2.5 and 4.5) run a blocking checklist of seven known AI-research failure modes, such as hallucinated results and citation errors. An opt-in claim audit can fetch a cited source and judge whether it supports the claim.
The README says this is not a fully autonomous AI scientist and not a tool for hiding AI use. It argues for human-led verification, and a Style Calibration feature learns your voice from past work. One caveat from the README: the claim audit has only been tested with a stub judge, and no live-judge calibration result is recorded yet.
What it does
Staged pipeline with confirmations
Work moves from research to writing, review, revision and finalization. The pipeline pauses for your confirmation at every stage.
Integrity gates
Stage 2.5 and Stage 4.5 run a 7-mode blocking checklist of AI research failure modes. Citations also carry source provenance and locator anchors.
Opt-in claim audit
Setting ARS_CLAIM_AUDIT=1 fetches each cited source and checks whether it supports the claim. Unsupported or fabricated references are blocked from output.
Style Calibration and writing checks
It learns your writing voice from past work and flags patterns that make prose read as machine-generated.
Several install routes
Install as a Claude Code plugin, as project or global skills, or import into Claude Science. Pi and Codex CLI versions are also mentioned.
What you could build
- Plan a paper's chapter structure through a Socratic dialogue with /ars-plan.
- Run a literature review on a topic with /ars-lit-review.
- Draft a paper in stages, confirming each step before moving on.
- Get a simulated peer review of your manuscript and revise against it.
- Check a draft's citations and claims against their cited sources before submitting.
Get started
Install Claude Code and set ANTHROPIC_API_KEY, then run the two plugin commands inside Claude Code. Try /ars-plan and describe your paper, or /ars-lit-review "your topic" for a quick single test.
/plugin marketplace add Imbad0202/academic-research-skills/plugin install academic-research-skillsCommands are copied from the project's README. Read it on GitHub for the full setup.
More in Agent skills & plugins
superpowers
A set of skills that makes coding agents plan, test and review before they write code
mattpocock/skills
Small, editable agent skills for planning, testing and reviewing code with AI
ECC
Agents, skills, hooks and memory that add an engineering workflow to coding agents
andrej-karpathy-skills
A CLAUDE.md file that tells Claude Code to ask first, keep code small, edit narrowly
anthropics/skills
Anthropic's example Agent Skills: folders of instructions Claude loads per task
ponytail
A skill that makes AI coding agents write the least code that works
gstack
Slash-command skills that give Claude Code roles like reviewer, QA and release engineer
addyosmani/agent-skills
A pack of 25 Markdown skills that guide AI coding agents from spec to ship
taste-skill
Agent skills that steer AI coding tools toward better layout, type and motion
archify
Agent skill that turns a description or a codebase into an interactive HTML diagram