К основному содержанию
плагин CX Codex

research-agent-skills — плагин для Claude Code и Codex

Плагин research-agent-skills добавляет 281 скилл для научных исследований в Claude Code и Codex. Охватывает весь цикл: от поиска литературы и выдвижения гипотез до проверки цитат и ответов рецензентам.

MIT ★ 1 596 обновлён 2026-10-29

Установить в Claude Code · Через маркетплейс плагинов Все способы и платформы ↓
/plugin marketplace add hashgraph-online/awesome-codex-plugins
/plugin install research-agent-skills@awesome-codex-plugins
Открыть на GitHub

Что делает research-agent-skills

Плагин работает на Claude Code и Codex, а также на Cursor, Gemini CLI, GitHub Copilot, Windsurf и OpenCode. Каждый скилл — отдельная папка с файлом SKILL.md, справочными материалами и проверенными скриптами. Агент загружает скилл только когда задача его требует, поэтому утром можно искать литературу, а вечером — готовить ответ рецензентам.

Всего 281 скилл, распределённых по этапам: генерация и проверка гипотез, поиск по базам данных, PRISMA-обзоры, планирование эксперимента и расчёт мощности, анализ данных через scanpy, rdkit, astropy, pydicom, pytorch-lightning и ещё 150+ инструментов, научный рейтинг и визуализация, подготовка к подаче в журнал или на arXiv, рецензирование и написание заявок на гранты.

Два скилла закрывают типичные провалы ИИ-помощников: unslop-academic-writing находит шаблонные обороты и монотонный ритм и переписывает текст в авторском голосе, а citation-verification проверяет каждую ссылку — потому что галлюцинированные цитаты в академическом тексте стоят дороже всего.

Какие задачи решает

  • провести систематический обзор литературы по протоколу PRISMA
  • сгенерировать и критически проверить гипотезы на пробелы
  • спланировать эксперимент и рассчитать размер выборки
  • проверить каждую цитату перед подачей рукописи
  • отформатировать статью под шаблон журнала или arXiv

Примеры запросов агенту

  • Проведи PRISMA-обзор по теме CRISPR off-target effects и сформируй таблицу включений
  • Запусти citation-verification для раздела References в manuscript.tex
  • Сгенерируй гипотезы по результатам в results.csv и укажи возможные пробелы

Как установить research-agent-skills

Через маркетплейс плагинов

/plugin marketplace add hashgraph-online/awesome-codex-plugins
/plugin install research-agent-skills@awesome-codex-plugins

По шагам, с командами для каждой платформы: Плагины Claude Code: маркетплейсы, установка и как собрать свой

Вопросы о research-agent-skills

Как установить research-agent-skills в Claude Code?

Выполните две команды: сначала /plugin marketplace add hashgraph-online/awesome-codex-plugins, затем /plugin install research-agent-skills@awesome-codex-plugins.

Что делает плагин research-agent-skills?

Добавляет 281 скилл для всего цикла научного исследования: от поиска литературы и выдвижения гипотез до анализа данных, написания статьи, проверки цитат и подготовки к рецензированию.

На каких платформах работает research-agent-skills?

На Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf и OpenCode — на любом хосте, поддерживающем Agent Skills.

Исходный текст

research-agent-skills · MIT
<div align="center">

<img src="PUBLIC/research-agent-skills-banner.webp" alt="Research Agent Skills for academia and scientific research by Kalaris Labs" width="100%" />

# Research Agent Skills

### Turn any AI agent into a research co-scientist

**<!-- count -->280+<!-- /count --> open-source agent skills that take Claude Code, Codex, Cursor, Gemini CLI, Copilot or any Agent Skills harness from the research question to the accepted paper: literature, hypotheses, experimental design, analysis in AI, ML, biology, chemistry, medicine and physics, writing, citation checks and peer review.**

[![skills.sh installs](https://img.shields.io/endpoint?url=https%3A%2F%2Fskills.sh%2Fapi%2Fbadge%2Fkalarislabs%2Fresearch-agent-skills&label=skills.sh%20installs&color=165d74)](https://skills.sh/kalarislabs/research-agent-skills)
[![Skills](https://img.shields.io/badge/skills-281-165d74)](#skill-catalog)
[![GitHub stars](https://img.shields.io/github/stars/KalarisLabs/research-agent-skills?style=flat&logo=github&label=stars&color=165d74)](https://github.com/KalarisLabs/research-agent-skills/stargazers)
[![README views](https://hits.sh/github.com/KalarisLabs/research-agent-skills.svg?label=views&color=165d74)](https://hits.sh/github.com/KalarisLabs/research-agent-skills/)
[![Last commit](https://img.shields.io/github/last-commit/KalarisLabs/research-agent-skills?color=165d74)](https://github.com/KalarisLabs/research-agent-skills/commits/main)
[![License: MIT](https://img.shields.io/badge/license-MIT-165d74)](LICENSE)

[![Validate](https://github.com/KalarisLabs/research-agent-skills/actions/workflows/validate.yml/badge.svg)](https://github.com/KalarisLabs/research-agent-skills/actions/workflows/validate.yml)
[![Skill security](https://github.com/KalarisLabs/research-agent-skills/actions/workflows/skill-security.yml/badge.svg)](https://github.com/KalarisLabs/research-agent-skills/actions/workflows/skill-security.yml)
[![Benchmarks](https://github.com/KalarisLabs/research-agent-skills/actions/workflows/benchmarks.yml/badge.svg)](benchmarks/README.md)
[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/KalarisLabs/research-agent-skills/badge)](https://securityscorecards.dev/viewer/?uri=github.com/KalarisLabs/research-agent-skills)
[![OpenSSF Best Practices](https://www.bestpractices.dev/projects/15771/badge)](https://www.bestpractices.dev/projects/15771)

**Works with**

[![Claude Code](https://img.shields.io/badge/Claude_Code-d97757?logo=claude&logoColor=white)](#install)
[![OpenAI Codex](https://img.shields.io/badge/OpenAI_Codex-111111)](#install)
[![Cursor](https://img.shields.io/badge/Cursor-222222?logo=cursor&logoColor=white)](#install)
[![Gemini CLI](https://img.shields.io/badge/Gemini_CLI-1a73e8?logo=googlegemini&logoColor=white)](#install)
[![GitHub Copilot](https://img.shields.io/badge/GitHub_Copilot-24292f?logo=githubcopilot&logoColor=white)](#install)
[![Windsurf](https://img.shields.io/badge/Windsurf-0b100f?logo=windsurf&logoColor=white)](#install)
[![OpenCode](https://img.shields.io/badge/OpenCode-333333)](#install)
[![Any Agent Skills harness](https://img.shields.io/badge/%2B_any_Agent_Skills_harness-165d74)](#works-in-any-harness)

```bash
npx skills add KalarisLabs/research-agent-skills
```

<sub>One command via <a href="https://skills.sh/kalarislabs/research-agent-skills">skills.sh</a> · no account · pick all skills or only the ones you need · not on npm</sub>

[Install](#install) · [Proof](#proof-a-whole-brain-study-run-end-to-end-with-these-skills) · [Who it's for](#built-for-researchers) · [Benchmarks](#measured-not-claimed) · [Skills](#skill-catalog) · [skills.sh](https://skills.sh/kalarislabs/research-agent-skills) · [Docs](https://docs.kalarislabs.com/research-agent-skills) · [FAQ](#faq)

</div>

---

**Research Agent Skills**, maintained by **[Kalaris Labs](https://github.com/KalarisLabs)**, is a free, open-source library of
[agent skills](https://agentskills.io) that gives your AI agent the methods of a careful scientist. Each skill is a folder
with a `SKILL.md` playbook, reference material and tested scripts. Your agent loads one only when the task needs it, so
the same agent can search the literature in the morning, design an experiment at noon and answer reviewers at night.

### From assistant to co-scientist

An assistant drafts what you ask for. A co-scientist works through the research itself and checks its own output.
The skills cover every stage:

| Stage | Your agent can | Skills |
|---|---|---|
| **Question** | Generate and stress-test hypotheses, find gaps | [`hypothesis-generation`](skills/hypothesis-generation/SKILL.md), [`scientific-brainstorming`](skills/scientific-brainstorming/SKILL.md), [`scientific-critical-thinking`](skills/scientific-critical-thinking/SKILL.md) |
| **Literature** | Search papers and databases, run PRISMA reviews | [`literature-review`](skills/literature-review/SKILL.md), [`paper-lookup`](skills/paper-lookup/SKILL.md), [`systematic-review-prisma`](skills/systematic-review-prisma/SKILL.md), [`database-lookup`](skills/database-lookup/SKILL.md) |
| **Design** | Plan experiments, size samples, pre-register | [`experimental-design`](skills/experimental-design/SKILL.md), [`statistical-power`](skills/statistical-power/SKILL.md) |
| **Analysis** | Analyze data with field-standard tools | [`statistical-analysis`](skills/statistical-analysis/SKILL.md), [`scanpy`](skills/scanpy/SKILL.md), [`rdkit`](skills/rdkit/SKILL.md), [`astropy`](skills/astropy/SKILL.md), [`pydicom`](skills/pydicom/SKILL.md), [`pytorch-lightning`](skills/pytorch-lightning/SKILL.md) and 150+ more |
| **Writing** | Draft, visualize and remove AI slop | [`scientific-writing`](skills/scientific-writing/SKILL.md), [`scientific-visualization`](skills/scientific-visualization/SKILL.md), [`unslop-academic-writing`](skills/unslop-academic-writing/SKILL.md) |
| **Verification** | Check every reference and reproducibility claim | [`citation-verification`](skills/citation-verification/SKILL.md), [`reproducibility-statement`](skills/reproducibility-statement/SKILL.md) |
| **Submission** | Format for the venue, prepare arXiv, write grants | [`venue-templates`](skills/venue-templates/SKILL.md), [`arxiv-submission`](skills/arxiv-submission/SKILL.md), [`research-grants`](skills/research-grants/SKILL.md) |
| **Review** | Review papers and answer reviewers | [`peer-review`](skills/peer-review/SKILL.md), [`rebuttal-and-response-to-reviewers`](skills/rebuttal-and-response-to-reviewers/SKILL.md) |
| **Autonomy** | Run supervised, multi-step research loops | [`autoresearch`](skills/autoresearch/SKILL.md) |

It is built against the three ways AI research help goes wrong:

1. **AI slop.** `unslop-academic-writing` finds stock phrasing ("delves into", "pivotal role"), empty emphasis, stacked hedges and
   monotone rhythm, then rewrites toward specific, evidence-backed prose in the author's own voice.
2. **Hallucinated citations.** `citation-verification` checks every reference against Crossref, OpenAlex and arXiv. On our labeled
   benchmark it flagged every fabricated and corrupted reference and passed every correct one.
3. **Outdated venue rules.** Journal skills make the agent read the current author guidelines before formatting, instead of trusting memory.

### How it works

```mermaid
flowchart LR
    A["You ask:<br/>'Check every reference in refs.bib'"] --> B["Your agent<br/>(Claude Code, Codex, Cursor…)"]
    B -- "reads short skill descriptions" --> C{"Which skill fits?"}
    C -- "loads only that one" --> D["citation-verification<br/>SKILL.md + scripts"]
    D -- "queries" --> E[("Crossref · OpenAlex · arXiv")]
    E --> F["Report: verified,<br/>mismatched, not found"]
```

Installed skills cost almost nothing until used: the agent sees one short description per skill and opens the
full playbook only for the task at hand.

## Proof: a whole-brain study, run end to end with these skills

<a href="https://github.com/KalarisLabs/fly-brain-resuscitation"><img src="https://raw.githubusercontent.com/KalarisLabs/fly-brain-resuscitation/main/assets/banner.png" alt="Simulated adult fly brain (FlyWire FAFB v783) during the reperfusion runaway, from the fly-brain-resuscitation study" width="100%" /></a>

**[Reviving a model fly brain](https://github.com/KalarisLabs/fly-brain-resuscitation)** is a Kalaris Labs preprint on anoxia, brain death and resuscitation in a
connectome-constrained simulation of the whole adult *Drosophila* brain (139,255 neurons). Claude worked through
Research Agent Skills and Kalaris Labs Myriad from the first hypothesis to the finished paper, under the author's
direction (disclosed in the paper; Claude is not an author).

| Stage | What the agent produced | How the skills shaped it |
|---|---|---|
| **Question** | 6 preregistered hypotheses, each with rival explanations, discriminating predictions and falsification thresholds, frozen before any simulation ran | [`hypothesis-generation`](skills/hypothesis-generation/SKILL.md) v2.2 workflow, cited in the [preregistration](https://github.com/KalarisLabs/fly-brain-resuscitation/blob/main/docs/HYPOTHESES.md) |
| **Literature** | 82 Crossref-verified sources and an 81-row parameter table, with 6 values flagged `UNVERIFIED` instead of guessed | [`citation-verification`](skills/citation-verification/SKILL.md) rule: 2 DOIs that resolved to the wrong papers were caught and discarded ([bibliography](https://github.com/KalarisLabs/fly-brain-resuscitation/blob/main/docs/lit/bibliography.md)) |
| **Experiments** | 3,487 analysed condition-seed simulation runs: network structure in 4 connectomes (FAFB, BANC, MANC, MCNS), dynamics in 2 (FAFB, MCNS) | Preregistered analysis plan; graded reperfusion (H3) was **falsified**, direction reversed, and reported as such |
| **Output** | 30-page paper, supplement, 19 figures, narrated explainer videos, unit-tested code | Every headline claim tied to its source results file, number and confidence level in [`KEY_CLAIMS.json`](https://github.com/KalarisLabs/fly-brain-resuscitation/blob/main/paper/KEY_CLAIMS.json) |

<table>
<tr>
<td width="50%" align="center"><a href="https://github.com/KalarisLabs/fly-brain-resuscitation"><img src="https://raw.githubusercontent.com/KalarisLabs/fly-brain-resuscitation/main/assets/C3_runaway.gif" width="100%" alt="Simulation: abrupt reperfusion after anoxia drives the model brain into a potassium-coupled runaway" /></a><br/><sub><b>Abrupt reoxygenation:</b> after 180 s of anoxia the model is trapped in a self-sustained K⁺-coupled high-activity state (48/48 FAFB parameter sets, 5/5 MCNS runs; confidence high)</sub></td>
<td width="50%" align="center"><a href="https://github.com/KalarisLabs/fly-brain-resuscitation"><img src="https://raw.githubusercontent.com/KalarisLabs/fly-brain-resuscitation/main/assets/C4_reset.gif" width="100%" alt="Simulation: a 20-second total activity block at reperfusion resets the network and its responses return" /></a><br/><sub><b>20-second total activity block at reperfusion:</b> the stimulus-response map returns (FAFB FII<sub>norm</sub> 1.001 / 0.953 at 30 / 180 s anoxia; confidence high)</sub></td>
</tr>
</table>

<sub><b>Scope:</b> all results are predictions of a simplified simulation (connectome-constrained LIF neurons and
phenomenological metabolism, constants not fitted to <i>Drosophila</i>). They are not evidence about living tissue,
clinical brain death or human resuscitation, and resuscitation of a brain-dead organism has not been demonstrated.
Claims and numbers follow the study's <a href="https://github.com/KalarisLabs/fly-brain-resuscitation/blob/main/paper/KEY_CLAIMS.json">KEY_CLAIMS.json</a>;
see also its <a href="https://github.com/KalarisLabs/fly-brain-resuscitation/blob/main/docs/LIMITATIONS.md">limitations</a>.</sub>

## Install

One command with the open [skills.sh](https://skills.sh/kalarislabs/research-agent-skills) CLI. It needs Node.js 18+
and no account, and works with the agents above plus about 70 others (Cline, Goose, Roo, Zed, Warp and more).

```bash
npx skills add KalarisLabs/research-agent-skills
```

That opens a picker for skills and agents. To skip the picker:

| I want to… | Run |
|---|---|
| **Install every skill** for every agent I use | `npx skills add KalarisLabs/research-agent-skills --all` |
| **Install specific skills** | `npx skills add KalarisLabs/research-agent-skills --skill citation-verification apa7` |
| See every skill before choosing | `npx skills add KalarisLabs/research-agent-skills --list` |
| Install for one agent only | add `--agent claude-code` (or `codex`, `cursor`, `gemini-cli`, `github-copilot`, …) |
| Install for all my projects | add `--global` (default is the current project) |
| Skip confirmation prompts | add `--yes` |
| Update installed skills later | `npx skills update` |

Then ask your agent in plain language, for example *"Verify every reference in refs.bib"* or
*"Unslop this introduction without changing any claims"*. The agent loads the right skill on its own.

Choose a starting skill for your field:

| Researcher | Start with |
|---|---|
| Academia across disciplines | `npx skills add KalarisLabs/research-agent-skills --skill scientific-writing` |
| Machine learning | `npx skills add KalarisLabs/research-agent-skills --skill ml-paper-writing` |
| Artificial intelligence | `npx skills add KalarisLabs/research-agent-skills --skill evaluating-llms-harness` |
| Biology and bioinformatics | `npx skills add KalarisLabs/research-agent-skills --skill scanpy` |
| Chemistry and materials | `npx skills add KalarisLabs/research-agent-skills --skill rdkit` |
| Medicine and clinical research | `npx skills add KalarisLabs/research-agent-skills --skill clinical-reports` |
| Physics and astronomy | `npx skills add KalarisLabs/research-agent-skills --skill astropy` |

**Field bundles in Claude Code.** The repository is also a Claude Code plugin marketplace with curated bundles:

```text
/plugin marketplace add KalarisLabs/research-agent-skills
/plugin install biology-research@research-agent-skills
```

Bundles: `research-essentials`, `ml-research`, `ai-research`, `biology-research`, `chemistry-research`,
`medicine-research`, `physics-research`, plus one plugin per category (for example `literature-review`).
[Browse all categories](catalog/skills.json) when you need a specialist skill.

<details>
<summary><b>Other installation methods</b></summary>

| Method | Command |
|---|---|
| GitHub CLI | `gh skill install KalarisLabs/research-agent-skills` |
| macOS / Linux, checksum-verified release | `curl -fsSL https://raw.githubusercontent.com/KalarisLabs/research-agent-skills/main/install.sh \| sh` |
| Windows, checksum-verified release | `irm https://raw.githubusercontent.com/KalarisLabs/research-agent-skills/main/install.ps1 \| iex` |
| A field bundle (standalone, no Node.js needed) | `curl -fsSL https://raw.githubusercontent.com/KalarisLabs/research-agent-skills/main/install.sh \| RAS_BUNDLE=biology-research sh` |

The standalone installers verify pinned release downloads with SHA-256. Full guide: [Installation](docs/getting-started/installation.md).

**Not on npm.** There is no `research-agent-skills` package on the npm registry. `npx skills` is the
open [skills.sh](https://skills.sh) CLI, which reads this GitHub repository directly. Any npm package
claiming to be this project is not ours.

</details>

### Works in any harness

The skills follow the open [Agent Skills specification](https://agentskills.io/specification): plain folders with a
`SKILL.md` file, no SDK, no server, no lock-in. Every skill passes the specification's reference validator.

| Harness | How |
|---|---|
| Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, Windsurf, Cline, Goose, Amp, Roo, Zed, Warp and about 60 more | `npx skills add KalarisLabs/research-agent-skills --agent <name>` |
| Claude Code plugin marketplace | `/plugin marketplace add KalarisLabs/research-agent-skills` |
| Any other harness that reads skills | Copy `skills/<name>/` into its skills directory (often `.agents/skills/`) |
| Agents without skill support | Point the agent at `skills/<name>/SKILL.md`; it is plain Markdown with scripts beside it |

Tool permissions and advanced metadata may vary by harness.

## Built for researchers

| You are... | Start with |
|---|---|
| **A PhD or master's student** writing a thesis, dissertation or first paper | `scientific-writing`, `unslop-academic-writing`, `literature-review`, `apa7` or your journal's skill · [Thesis guide](docs/guides/thesis-and-dissertation.md) |
| **A researcher or PI** submitting manuscripts and grants | journal-format skills, `abstract-and-title`, `cover-letter-to-editor`, `rebuttal-and-response-to-reviewers`, `research-grants` · [Paper guide](docs/guides/write-a-research-paper.md) |
| **Running a systematic review or meta-analysis** | `systematic-review-prisma`, `citation-verification`, `statistical-analysis` · [Review guide](docs/guides/systematic-review.md) |
| **An ML / AI researcher** | `ml-paper-writing` (NeurIPS, ICML, ICLR, ACL, AAAI templates), `arxiv-submission`, and the `ml-*` categories |
| **In biology, chemistry, medicine, physics or earth science** | 100+ database and analysis skills · [Skills by field](docs/guides/by-field.md) |
| **A librarian or research software engineer** | `reference-manager-interop` (Zotero, Mendeley, EndNote), `paper-corpus-rag`, `research-knowledge-graph`, `research-skill-creator` |

### Find an agent skill for your research task

| Research task | Start here |
|---|---|
| Search papers and build a literature review | [`literature-review`](skills/literature-review/SKILL.md), [`research-lookup`](skills/research-lookup/SKILL.md), [literature-review skills](catalog/lists/literature-review.txt) |
| Discover papers in Firecrawl's hosted Research Index | [`firecrawl-research-index`](skills/firecrawl-research-index/SKILL.md) searches indexed papers, reads relevant passages, and follows citations; access is currently free with limits ([provider docs](https://docs.firecrawl.dev/features/research)) |
| Plan and report a systematic review or meta-analysis | [`systematic-review-prisma`](skills/systematic-review-prisma/SKILL.md), [`statistical-analysis`](skills/statistical-analysis/SKILL.md), [systematic review guide](docs/guides/systematic-review.md) |
| Draft or revise a research paper, thesis or grant | [`scientific-writing`](skills/scientific-writing/SKILL.md), [`unslop-academic-writing`](skills/unslop-academic-writing/SKILL.md), [`research-grants`](skills/research-grants/SKILL.md), [writing guide](docs/guides/write-a-research-paper.md) |
| Check citations, BibTeX and reference libraries | [`citation-verification`](skills/citation-verification/SKILL.md), [`bibtex-hygiene`](skills/bibtex-hygiene/SKILL.md), [`reference-manager-interop`](skills/reference-manager-interop/SKILL.md) |
| Prepare a journal or conference submission | [journal format skills](catalog/lists/journal-formats.txt), [`arxiv-submission`](skills/arxiv-submission/SKILL.md), [`rebuttal-and-response-to-reviewers`](skills/rebuttal-and-response-to-reviewers/SKILL.md) |
| Analyze data and report reproducible methods | [data science skills](catalog/lists/data-science-and-ml.txt), [`scientific-visualization`](skills/scientific-visualization/SKILL.md), [`reproducibility-statement`](skills/reproducibility-statement/SKILL.md) |

### Explore scientific agent skills by discipline

| Field | Example skills | Browse |
|---|---|---|
| Bioinformatics and genomics | [`scanpy`](skills/scanpy/SKILL.md), [`scvi-tools`](skills/scvi-tools/SKILL.md), [`pydeseq2`](skills/pydeseq2/SKILL.md) | [Life sciences](catalog/lists/life-sciences.txt) |
| Chemistry, materials and drug discovery | [`rdkit`](skills/rdkit/SKILL.md), [`deepchem`](skills/deepchem/SKILL.md), [`pymatgen`](skills/pymatgen/SKILL.md) | [Chemistry and drug discovery](catalog/lists/chemistry-and-drug-discovery.txt) |
| Clinical, biomedical and health research | [`pydicom`](skills/pydicom/SKILL.md), [`pyhealth`](skills/pyhealth/SKILL.md), [`clinical-reports`](skills/clinical-reports/SKILL.md) | [Clinical and health](catalog/lists/clinical-and-health.txt) |
| Physics, astronomy and earth science | [`astropy`](skills/astropy/SKILL.md), [`qiskit`](skills/qiskit/SKILL.md), [`geopandas`](skills/geopandas/SKILL.md) | [Physical sciences](catalog/lists/physical-sciences.txt) |
| Psychology and social science | [`apa7`](skills/apa7/SKILL.md), [`experimental-design`](skills/experimental-design/SKILL.md), [`statistical-power`](skills/statistical-power/SKILL.md) | [Research essentials](catalog/lists/research-essentials.txt) |
| AI and machine learning research | [`ml-paper-writing`](skills/ml-paper-writing/SKILL.md), [`pytorch-fsdp2`](skills/pytorch-fsdp2/SKILL.md), [`evaluating-llms-harness`](skills/evaluating-llms-harness/SKILL.md) | [ML training](catalog/lists/ml-training.txt) |

For more workflows and installation choices, see [skills by field](docs/guides/by-field.md).

## What you can ask

```text
"Draft the Results section from results/ and the figures in figs/, then unslop it."
"Verify every reference in refs.bib and tell me which ones don't exist."
"Reformat this manuscript for Cell: Summary, Highlights, eTOC blurb and STAR Methods."
"Run a PRISMA 2020 systematic review on exercise and depression: search strings, dedup, flow diagram."
"Write a point-by-point response to these three reviewers within the rebuttal limit."
"Build a citation graph of my Zotero library and find the papers that bridge the two clusters."
"Check my LaTeX source before I upload it to arXiv."
"Analyze this scRNA-seq dataset with scanpy and write the methods section."
```

## Measured, not claimed

Every skill is scored on a static quality rubric and a trigger-routing suite, and task outcomes are measured for the
original skills. Every chart below is rendered from the committed results in
[`benchmarks/results/`](benchmarks/results/) by [`benchmarks/charts.py`](benchmarks/charts.py), and CI fails if a chart
drifts from its data. Methodology and caveats: [benchmarks/README.md](benchmarks/README.md).

<picture><source media="(prefers-color-scheme: dark)" srcset="PUBLIC/benchmarks/task-outcomes-dark.svg" /><img src="PUBLIC/benchmarks/task-outcomes-light.svg" alt="Dumbbell chart: judge score per task with and without the skill. The skill wins 21 of 23 tasks, mean 7.4 versus 4.3." width="100%" /></picture>

<table>
<tr>
<td width="50%"><picture><source media="(prefers-color-scheme: dark)" srcset="PUBLIC/benchmarks/slop-dark.svg" /><img src="PUBLIC/benchmarks/slop-light.svg" alt="Bar chart: mean slop index 0.8 with the unslop skill, 5.0 without it, 2.9 for human abstracts. 83 percent lower with the skill." width="100%" /></picture></td>
<td width="50%"><picture><source media="(prefers-color-scheme: dark)" srcset="PUBLIC/benchmarks/citations-dark.svg" /><img src="PUBLIC/benchmarks/citations-light.svg" alt="Bar chart: citation-verification passed 10 of 10 real references and flagged 4 of 4 corrupted and 4 of 4 fabricated ones." width="100%" /></picture></td>
</tr>
<tr>
<td width="50%"><picture><source media="(prefers-color-scheme: dark)" srcset="PUBLIC/benchmarks/routing-dark.svg" /><img src="PUBLIC/benchmarks/routing-light.svg" alt="Grouped bars: on 100 held-out prompts an LLM router picks the right skill hit@3 0.96 with 0.00 false triggers, versus 0.77 and 0.04 for keyword routing." width="100%" /></picture></td>
<td width="50%"><picture><source media="(prefers-color-scheme: dark)" srcset="PUBLIC/benchmarks/quality-dark.svg" /><img src="PUBLIC/benchmarks/quality-light.svg" alt="Histogram of static quality scores: all 281 skills score 80 or more, mean 95.2." width="100%" /></picture></td>
</tr>
</table>

<details>
<summary><b>The same numbers as a table</b></summary>

| Check | Result |
|---|---|
| Static quality rubric (structure, triggers, verification rules, tests) | Original skills: mean **95.5/100** · all skills: 95.3 |
| Trigger routing: 50 reviewed prompts, including near misses | hit@3 **1.00**, hit@1 1.00; false-trigger 0.12 |
| Citation verification on 18 labeled references (real, corrupted, fabricated) | precision **1.00**, recall **1.00** |
| No-slop writing: model abstracts with vs without `unslop-academic-writing` (pilot) | slop index **83% lower**, clean share 68% → 89% |
| Task outcomes with vs without skill, blind LLM judge (pilot, 23 tasks) | skill wins **21/23**, mean score 7.4 vs 4.3 |

</details>

Model-based results are pilots (small samples, one small model as answerer and judge). Quality and routing gates run
on every pull request, so a weak or overlapping skill can't be merged silently.

## Featured skills

| Skill | What it does |
|---|---|
| [`unslop-academic-writing`](skills/unslop-academic-writing/SKILL.md) | Removes AI slop from papers, theses and grants: linter plus rewrite playbook, in your voice |
| [`citation-verification`](skills/citation-verification/SKILL.md) | Detects fabricated or mismatched references via Crossref, OpenAlex and arXiv |
| [`scientific-writing`](skills/scientific-writing/SKILL.md) | IMRaD manuscripts for any field, with reporting guidelines (CONSORT, STROBE, PRISMA) |
| [`ml-paper-writing`](skills/ml-paper-writing/SKILL.md) | Publication-ready ML papers with official NeurIPS, ICML, ICLR, ACL, AAAI and COLM LaTeX templates |
| [`systematic-review-prisma`](skills/systematic-review-prisma/SKILL.md) | Protocol → search → dedup → screening → PRISMA 2020 flow diagram and checklist |
| [`rebuttal-and-response-to-reviewers`](skills/rebuttal-and-response-to-reviewers/SKILL.md) | Triage reviews, then write journal responses or length-limited conference rebuttals |
| [`abstract-and-title`](skills/abstract-and-title/SKILL.md) | Abstracts to word limits, titles, keywords, highlights, significance statements |
| [`venue-templates`](skills/venue-templates/SKILL.md) | Journal, conference, poster and grant templates with verification-first compliance |
| [`paper-corpus-rag`](skills/paper-corpus-rag/SKILL.md) | Ask questions over your own PDFs and get answers citing the exact passage |
| [`firecrawl-research-index`](skills/firecrawl-research-index/SKILL.md) | Discover arXiv and biomedical papers in Firecrawl's hosted index, inspect records, and read question-matched passages |
| [`research-knowledge-graph`](skills/research-knowledge-graph/SKILL.md) | Citation, author and topic graphs exported to NetworkX, Neo4j, Gephi, Obsidian and graphify |
| [`database-lookup`](skills/database-lookup/SKILL.md) | One interface to 100+ public scientific databases |
| [`research-skill-creator`](skills/research-skill-creator/SKILL.md) | Turn your lab's workflows into tested, secure skills |

## Journal and venue formats

Nature · Nature Communications · Scientific Reports · Science · Science Advances · Cell · Neuron · Cell Reports ·
IEEE Transactions · IEEE Access · ACM (sigconf, acmsmall: CHI, KDD, SIGGRAPH) · Elsevier (elsarticle, CAS) ·
Springer LNCS · PLOS ONE · PLOS Biology · APA 7th edition · arXiv · NeurIPS · ICML · ICLR · ACL/ARR · AAAI ·
COLM · OSDI/NSDI/ASPLOS/SOSP · NIH and NSF grants · conference posters and slides.

## Skill catalog

Grouped by category and generated from [`catalog/skills.json`](catalog/skills.json). The
[documentation directory](docs/skills/) has a page per skill.

<!-- catalog:start -->

<details><summary><b>research-writing</b> (15) · Scientific and research paper writing, grants, peer review, critical appraisal</summary>

| Skill | What it does |
|---|---|
| [`abstract-and-title`](skills/abstract-and-title/SKILL.md) | Write and sharpen research paper titles, abstracts (structured and unstructured), keywords, highlights, significance statements, graphical-abstract text and la… |
| [`cover-letter-to-editor`](skills/cover-letter-to-editor/SKILL.md) | Write journal submission cover letters, presubmission inquiries, transfer requests, and reviewer suggestion or exclusion lists that help a manuscript get past… |
| [`dhdna-profiler`](skills/dhdna-profiler/SKILL.md) | Extract cognitive patterns and thinking fingerprints from any text. |
| [`markdown-mermaid-writing`](skills/markdown-mermaid-writing/SKILL.md) | Writes scientific documents and documentation as markdown with embedded Mermaid diagrams as the canonical, git-diffable source format. |
| [`market-research-reports`](skills/market-research-reports/SKILL.md) | Build evidence-traceable market research reports and assumption-driven market sizing or forecast scenarios. |
| [`ml-paper-writing`](skills/ml-paper-writing/SKILL.md) | Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. |
| [`peer-review`](skills/peer-review/SKILL.md) | Prepare evidence-bounded, constructive peer-review drafts and structured manuscript assessments. |
| [`rebuttal-and-response-to-reviewers`](skills/rebuttal-and-response-to-reviewers/SKILL.md) | Plan and write responses to peer review, including journal "response to reviewers" letters for revise-and-resubmit, conference rebuttals under strict length li… |
| [`reproducibility-statement`](skills/reproducibility-statement/SKILL.md) | Prepare the reproducibility, transparency and open-science parts of a paper, including data and code availability statements, reproducibility checklists (NeurI… |
| [`research-grants`](skills/research-grants/SKILL.md) | Guides writing of competitive research grant proposals for NSF, NIH, DOE, DARPA, and Taiwan NSTC. |
| [`scholar-evaluation`](skills/scholar-evaluation/SKILL.md) | Provide qualitative-first, evidence-traceable developmental review of scholarly works and audit low-stakes research-assessment rubrics with optional local qual… |
| [`scientific-critical-thinking`](skills/scientific-critical-thinking/SKILL.md) | Evaluate scientific claims and evidence quality. |
| [`scientific-writing`](skills/scientific-writing/SKILL.md) | Draft, revise, and audit scientific manuscripts or reports with explicit evidence provenance, r...

© Kalaris Labs · MIT · hashgraph-online/awesome-codex-plugins

Похожие плагины

Все плагины по теме «Наука и исследования» →
academic-research-suite Imbad0202 Скилл academic-research-suite маршрутизирует задачи академического пайплайна: от обзора литературы и проверки цитирования до симуляции рецензирования. Запускается по слэш-командам и ключевым фразам. CX Codex +1 2 200установок research-writing-literature hanhuark Скилл research-writing-literature помогает выстраивать строгую научную аргументацию, готовить обзоры литературы, описывать методы и разбирать графики в статьях. CX Codex +1 ★ 26звёзд репо research-survey EvoScientist Скилл research-survey собирает из найденных статей структурированный обзор поля: план, черновик, расширение по разделам, сравнительные таблицы и ссылки. CX Codex +1 318установок research-craft curiositech Скилл research-craft для Claude Code помогает формулировать научные проблемы, проверять гипотезы и структурировать диссертации или статьи. CX Codex +1 153установок academic-research akillness Скилл academic-research маршрутизирует запрос по теме исследования или тексту статьи в одну из четырёх смежных методик — от поиска литературы до рецензирования. Решение, какую ветку и режим запускать, принимает сам скилл. CX Codex +1 148установок academic-deep-research kesslerio Скилл academic-deep-research для Claude Code проводит многошаговые исследования: два цикла поиска на тему, иерархия доказательств, ссылки по APA 7. Создаёт обзоры литературы и конкурентную разведку. CX Codex +1 562установок

Из того же набора391

Весь набор →
runtype-skills Набор скиллов для разработки AI-продуктов на Runtype: от настройки проекта до инструментов агента, Persona-виджетов и рабочих процессов. CX Codex ★ 1 596звёзд репо roadmapsmith Плагин roadmapsmith держит ROADMAP.md в согласии с кодом: одна команда создаёт файл, вторая предлагает правки с подтверждением и ждёт вашего ok. CX Codex ★ 1 596звёзд репо river-review Плагин river-review запускает аудит кода по версионируемым правилам команды прямо в Claude Code и Codex с проверкой находок по diff и тестам. CX Codex ★ 1 596звёзд репо reviewable-html-workbench Плагин reviewable-html-workbench превращает HTML-документы агента в документы для браузерной проверки с inline-комментариями и повторными правками. Подходит для отчётов, дизайн-документов и таблиц сравнения. CX Codex ★ 1 596звёзд репо repopilot Плагин repopilot фиксирует срез репозитория на старте сессии и блокирует завершение работы агента, если тот ослабил тесты или убрал проверки безопасности. CX Codex ★ 1 596звёзд репо rekall Плагин rekall сжимает историю диалога в расширении Codex для VS Code, сохраняет письменную передачу задачи и продолжает работу один раз. CX Codex ★ 1 596звёзд репо registry-broker-codex-plugin Плагин registry-broker-codex-plugin отдаёт задачи Codex внешним агентам через HOL Registry Broker: сначала спрашивает брокера, потом показывает кандидатов или отправляет подзадачу. CX Codex ★ 1 596звёзд репо account-video-downloader Плагин account-video-downloader пакетно выгружает все видео с публичных страниц авторов на YouTube, TikTok, Douyin, Kuaishou и Bilibili без водяных знаков. CX Codex ★ 1 596звёзд репо

Спонсоры проекта

Компании, которые помогают держать аналитику открытой