Paper Skills overview

What It Is

Paper Skills is an agent-readable skill library for the work around research papers: drafting, polishing, citations, reviewer response, venue expectations, and scientific figures.

The project starts from a practical observation: every discipline has its own paper language. That language is not only prose. It includes evidence patterns, reference habits, review standards, figure grammar, and venue-specific taste.

Skill Areas

Area What it should cover Current state
writing Polishing, translation, restructuring, and argument repair. scaffold
citation Literature search, reference checks, citation audit, and exports. scaffold
review Reviewer simulation, rebuttal letters, revision QA, and response workflows. scaffold
venue-packs Journal-, conference-, and discipline-specific conventions. scaffold
plot Field-specific scientific figures and editable figure skills. seed cases

Why This Repository Exists

Many academic skill packs already solve parts of the workflow. Paper Skills is not trying to rebuild all of them from scratch. The first goal is to curate, acknowledge, and interoperate with strong public work.

The first first-party focus is plot: field-specific scientific figures. A materials figure, a medical figure, a traffic figure, an integrated-circuit schematic, and a computer-science system diagram should not share one generic diagram recipe. They need domain packs.

plot: Field-Specific Figures

Initial domain seeds:

Domain Path Figure families
Integrated circuits skills/plot/domains/integrated-circuits analog schematics, switched-capacitor figures, circuit architecture
Computer science skills/plot/domains/computer-science system diagrams, model architecture, pipelines, evaluation setup

Planned community directions include materials science, medicine and biomedicine, traffic and transportation, biology, and finance.

Contributors

Paper Skills credits contributors by the fields and work they contribute, not only by commit count.

Contributor Background Fields Contributions
Bubu @KoinaAI · CS PhD @ ZJU Computer science, scientific visualization, AI agents Creator, project direction, plot

Contributed a skill, domain pack, benchmark, review, or integration? Add your name, background, fields, and contribution to this table in the same pull request.

Repository Layout

skills/
  writing/
  citation/
  review/
  venue-packs/
  plot/
    domains/
      integrated-circuits/
      computer-science/
integrations/
  catalog.md
docs/
  academic-skill-pack-survey.md
registry.yaml

Install

No stable install path yet. This repository is still a public scaffold.

For now, use the skills by referencing the relevant SKILL.md or README files directly from an agent session.

Integration Policy

Paper Skills will prefer adapters and references before vendoring external skills. If upstream content is copied or modified later, it must follow the upstream license and preserve attribution.

See ACKNOWLEDGEMENTS.md.

Roadmap

  • Make plot demo cases concrete for integrated circuits and computer science.
  • Expand demos into reusable benchmark cases across all skill areas.
  • Expand the integration catalog with attribution notes and adapter plans.
  • Keep upstream attribution and contribution rules explicit as integrations land.
  • Add a stable install/update command after the first usable skill set.

Star History

License

MIT. External skill integrations still need their own license checks and attribution.