Skills

• PSresearch-repoboth

  ↗
  /oss:research-repo · $research-repo

  Set up a research project's folders the same way every time — sources, drafts, and a CLAUDE.md file the AI reads first — or check that an existing project already follows that layout.

• WOfable-orchestrateClaude only

  ↗
  /oss:fable-orchestrate

  Run a hard task past three AI models at once, with Fable 5 — the top model — as lead: Opus takes the deep reasoning, Sonnet the fast mechanical work, and Codex (GPT-5.6) the outside check, instead of trusting a single model's answer.

• WOopus-orchestrateClaude only

  ↗
  /oss:opus-orchestrate

  The same three-model teamwork as fable-orchestrate, but led by Claude Opus 4.8 running flat out — the lead does the hard thinking itself and hands off only parallel reasoning (Opus subagents), mechanical work (Sonnet), and the outside check (Codex, GPT-5.6).

• WO46-orchestrateCodex only

  ↗
  $46-orchestrate

  A Codex orchestrator — GPT-5.6 Sol at high effort owns the hard decisions, integration, and sign-off; it routes bounded work to Terra and reserves Luna for tightly specified mechanical work. The Codex counterpart to fable-orchestrate.

• WOadvisorboth

  ↗
  /oss:advisor · $advisor

  Ask for a second opinion before you commit to a decision or call work done. In Claude Code the advisor seat is always Fable 5, whatever model your session runs — a Sonnet session consults Fable, at the session's own effort level. In Codex, always Sol/high — the flagship tier, since the point is a stronger reviewer, not a cheap one. Advice only; it never edits your files.

• IDdivergeboth

  ↗
  /oss:diverge · $diverge

  Before committing to one approach, generate three to five genuinely different ones — safe, bold, unconventional — so you're choosing between real options instead of the first idea.

• IDdiverge-codexboth

  ↗
  /oss:diverge-codex · $diverge-codex

  The same brainstorming step as diverge, but handed to a second AI model (Codex) so the ideas don't all come from the same source.

• RDconjoint-designboth

  ↗
  /oss:conjoint-design · $conjoint-design

  Design a conjoint experiment — which attributes to test, how many respondents you need, and how you'll estimate each attribute's effect (AMCE/AMIE) — before you run it.

• RDconjoint-diagnosticsboth

  ↗
  /oss:conjoint-diagnostics · $conjoint-diagnostics

  Run through a checklist of what typically goes wrong in a conjoint experiment's design, execution, or analysis, and catch it before submission.

• RDconjoint-cleaningboth

  ↗
  /oss:conjoint-cleaning · $conjoint-cleaning

  Turn a raw conjoint survey export from Qualtrics into a clean, analysis-ready dataset.

• RDsurvey-designboth

  ↗
  /oss:survey-design · $survey-design

  Write survey questions and response scales that measure what you intend, flow logically, and don't quietly bias respondents toward a socially acceptable answer.

• RDcross-national-designboth

  ↗
  /oss:cross-national-design · $cross-national-design

  Adapt a survey experiment to run fairly across multiple countries — enough statistical power in each, localized wording, and a check for hidden bias between countries.

• RDlist-experimentboth

  ↗
  /oss:list-experiment · $list-experiment

  Design and check a list experiment (the item-count technique) — a way of asking about sensitive topics without anyone naming their own answer directly.

• ANtopic-modelingboth

  ↗
  /oss:topic-modeling · $topic-modeling

  Set up a structural topic model — group documents into themes and see how those themes relate to variables you care about — with the diagnostics to know if it worked.

• ANtext-classificationboth

  ↗
  /oss:text-classification · $text-classification

  Have an AI model sort text into categories you define, following a written codebook, and check its accuracy the way you'd check a human coder's.

• ANmodel-council-votingboth

  ↗
  /oss:model-council-voting · $model-council-voting

  Have several AI models independently label the same data, then combine their votes and measure how much they agreed — inter-rater reliability, but with models instead of people.

• ANmodel-committeeboth

  ↗
  /oss:model-committee · $model-committee

  Have two AI models (GPT-5.5 and Claude Opus) discuss a judgment call with each other before settling on an answer, instead of asking just one. This is the version where Claude Opus runs the meeting.

• ANmodel-committee-solboth

  ↗
  /oss:model-committee-sol · $model-committee-sol

  The same two-model committee as model-committee, but a third model — GPT-5.6 "Sol" — runs the meeting and makes the final call, so the tie-breaker isn't one of the two debaters.

• ANmodel-committee-fableboth

  ↗
  /oss:model-committee-fable · $model-committee-fable

  The same two-model committee as model-committee, but a fast, lightweight model (Fable 5) chairs — it tallies the debate and settles ties without adding an agenda of its own.

• ANllm-calibration-logprobsboth

  ↗
  /oss:llm-calibration-logprobs · $llm-calibration-logprobs

  Check how confident an AI model actually is in each answer, using the model's own token probabilities rather than just asking it to rate its confidence.

• CPvlm-ocr-pipelineboth

  ↗
  /oss:vlm-ocr-pipeline · $vlm-ocr-pipeline

  Design a pipeline that uses a vision-language model to read text out of scanned documents or images, instead of a traditional OCR engine.

• CPpost-ocr-cleanupboth

  ↗
  /oss:post-ocr-cleanup · $post-ocr-cleanup

  Clean up and quality-check text after it's been pulled out by OCR, catching the kinds of errors OCR typically introduces.

• CPvlm-ocr-evaluationboth

  ↗
  /oss:vlm-ocr-evaluation · $vlm-ocr-evaluation

  Before running OCR on a large batch of documents, compare a few candidate OCR systems on a small test set to see which is actually most accurate for your material.

• WRhypothesis-buildingboth

  ↗
  /oss:hypothesis-building · $hypothesis-building

  Turn a research idea into a hypothesis that could actually be proven wrong — with a causal diagram and a plan for testing it, including the possibility of no effect.

• WRliterature-reviewboth

  ↗
  /oss:literature-review · $literature-review

  Map out what's already been studied on a topic, spot the real gaps, and plan how to synthesize the literature you've found.

• WRnarrative-buildingboth

  ↗
  /oss:narrative-building · $narrative-building

  Draft or tighten a paper's introduction so the argument actually builds — why the question matters, and what follows if the answer goes one way or the other.

• WRpre-registration-writingboth

  ↗
  /oss:pre-registration-writing · $pre-registration-writing

  Write a pre-analysis plan: what you'll measure and how you'll analyze it, registered before you see the results.

• WRmethods-reportingboth

  ↗
  /oss:methods-reporting · $methods-reporting

  Check that your methods section reports everything expected by standard research-reporting checklists (CONSORT, JARS, DA-RT).

• WRpaper-texboth

  ↗
  /oss:paper-tex · $paper-tex

  Take a draft in any format — Word, Markdown, plain text — and typeset it into a properly formatted LaTeX paper, ready for a specific journal's requirements.

• FTfiguresboth

  ↗
  /oss:figures · $figures

  Design a publication-quality figure — chart type, color, scales, and captions chosen to make the data as easy to read as possible.

• FTtablesboth

  ↗
  /oss:tables · $tables

  Design a publication-quality table — column order, row grouping, number formatting, and what the notes underneath should say.

• MQfair-checkboth

  ↗
  /oss:fair-check · $fair-check

  Check whether a manuscript's data, code, and materials are actually available and usable by someone else — not just claimed to be.

• MQcitation-checkboth

  ↗
  /oss:citation-check · $citation-check

  Check every citation in a manuscript against its reference list, catching ones that don't exist, don't match, or are formatted wrong.

• MQfact-checkboth

  ↗
  /oss:fact-check · $fact-check

  Check whether the sources a manuscript cites actually say what the manuscript claims they say.

• MQfigure-table-auditboth

  ↗
  /oss:figure-table-audit · $figure-table-audit

  A final pass over a manuscript's figures and tables — captions, cross-references, and statistical notes — before submission.

• MQreplication-packageboth

  ↗
  /oss:replication-package · $replication-package

  Build or check a replication package — the folder of data, code, and documentation someone else would need to reproduce your results.

• RSpaper-review-liteboth

  ↗
  /oss:paper-review-lite · $paper-review-lite

  Run a manuscript through a pre-submission review that checks its claims against actual quotes from the text, catching overclaiming before a real reviewer does.

• RSpaper-review-lite-codexboth

  ↗
  /oss:paper-review-lite-codex · $paper-review-lite-codex

  The same pre-submission review as paper-review-lite, but run by two AI models independently, then cross-checked against each other.

• RSpresubmitClaude only

  ↗
  /oss:presubmit

  A guided walkthrough of a much larger, 30-plus-stage automated pre-submission review tool.

• RSjournal-reviewboth

  ↗
  /oss:journal-review · $journal-review

  Draft a referee report on someone else's manuscript, as if you were reviewing it for a journal.

No skills match your search. Try a different word, or clear the category filter.

This index mirrors Open Science Skills as of 11 July 2026 (v2.16.1). 39 skills on Claude Code (/oss:name) and 37 on Codex ($name); most run on both, a few are platform-specific.