Skills

Every scriptorium skill is single-responsibility, reads MANUSCRIPT_STATE.yaml, and emits structured markdown. The table below is generated from each skill’s manifest.yaml at docs-build time — category, lifecycle fit, modifies-manuscript, author-side-only, and required bibliography all come straight from the manifest fields.

Each row links to the skill’s README.md in the repo, which is the full operational contract (inputs, outputs, refusal behaviours, output schema, complete grounding list).

Categorisation axes

Three axes give you most of what you need to decide which skill to run next:

• Category — what kind of operation the skill performs.
  • critique — assesses; does not modify the manuscript.
  • validation — checks against an external standard; does not modify.
  • normalization — enforces declared style; suggests edits, never auto-applies.
  • transformation — modifies prose under a preservation contract.
  • meta — orientation or explanation; no manuscript modification.
  • utility — bootstrap; modifies only MANUSCRIPT_STATE.yaml.
• Lifecycle stage — which document_phase values the skill is invokable in. Skills refuse cleanly on phases for which they do not have enough declared state to ground against.
• Modifies the manuscript? — no (most critique skills), suggests (normalization and transformation skills emit diffs the author reviews and applies), or state file only (utility, or opt-in for some critique skills).

Two additional flags some authors need:

• Author-side only? — yes for skills whose manifest.yaml declares positioning: author-side-only. Editorial-side use violates ICMJE / NIH / major-publisher policy and the skill itself refuses to run on a manuscript the user did not author.
• Requires bibliography? — yes when the manifest’s bibliography input is required: true; optional when the input is declared but not required; no when the manifest does not declare a bibliography input at all.

All shipped skills

Skill	Category	Lifecycle fit	Modifies	Author-side only	Reqs bib
argumentative-flow	transformation	draft · review · revision	suggests	no	yes
author-contribution-audit	critique	review · revision · submission	no	no	no
citation-audit	critique	draft · review · revision · submission	no	no	yes
compression	transformation	revision · submission	suggests	no	yes
desk-rejection-risk	critique	revision · submission	no	yes	no
explain	meta	any phase	no	no	no
figure-text-alignment	critique	draft · review · revision	no	no	no
gap-finder	critique	draft · review · revision	no	no	no
init	utility	any phase	state file only	no	no
outlier-sentence-detector	critique	draft · review · revision · submission	no	no	no
reporting-guideline-compliance	validation	review · revision · submission	no	no	no
reporting-guideline-fit	critique	draft · review · revision	no	no	no
reviewer-simulation	critique	review · revision · submission	no	yes	yes
terminology-normalization	normalization	draft · review · revision	suggests	no	yes
tour	meta	any phase	no	no	no
venue-fit	critique	review · revision · submission	state file only (opt-in)	no	no

Per-category detail

Critique

Critique skills assess existing prose and emit structured findings. They do not modify the manuscript and they do not invent citations.

Skill	One line	Primary grounding
author-contribution-audit	Audit the manuscript’s Author Contributions section against ICMJE’s four authorship criteria and CRediT’s 14 contributor roles.	credit-taxonomy-authorship
citation-audit	Audit a manuscript’s existing citations for claim-support alignment, primary-vs-review mismatch, causal overreach, and unsupported assertions.	citation-claim-alignment, citation-accuracy-evidence, citation-overreach-research
desk-rejection-risk	Author-side pre-submission audit that flags triggers likely to result in desk rejection before peer review — scope/audience mismatch, format and length issues, missing or weak required sections, weak significance framing, and presentation patterns editors triage on.	editorial-decision-making, significance-positioning, common-critiques-taxonomy
figure-text-alignment	Audit text-only alignment between figure captions and the body-text sentences that reference them.	visualization-figures, internal-consistency, forensic-methodology
gap-finder	Identify gaps in existing draft prose (claims under-supported, premises missing, related work not engaged with, counterarguments not addressed, internal-consistency mismatches across sections) and suggest directions for filling them.	research-gap-detection, literature-search-strategies, argument-mapping
outlier-sentence-detector	Flag sentences that are statistical outliers in the manuscript along three narrow dimensions — length, syntactic complexity, and nominalization density — for human review.	quantitative-quality-measures, copyediting-vs-developmental
reporting-guideline-fit	Read the manuscript’s methods section and infer which EQUATOR Network reporting guideline applies — CONSORT, STROBE, PRISMA, ARRIVE, TRIPOD/TRIPOD+AI, STARD, CARE, COREQ, CHEERS, plus AI-extensions where applicable.	reporting-guidelines
reviewer-simulation	Author-side simulation of peer review across four attentional lenses (methodological skeptic, domain expert, translational/clinical, statistical).	reviewer-archetypes-evidence, common-critiques-taxonomy, ai-peer-review-research
venue-fit	Assess venue fit for a manuscript and return a tiered set of candidate venues (likely fit / stretch / probably premature) with per-axis rationale, explicit predatory-venue refusal, and opt-in preprint server + pre/post-publication review platform recommendations.	venue-selection, predatory-publishing, preprint-landscape

Validation

Validation skills check existing prose against an external standard and emit structured findings. They do not modify the manuscript.

Skill	One line	Primary grounding
reporting-guideline-compliance	Run an EQUATOR Network reporting-guideline checklist (CONSORT, STROBE, PRISMA, ARRIVE, STARD, TRIPOD/TRIPOD+AI, CARE, COREQ, CHEERS, plus AI extensions) against a manuscript.	reporting-guidelines, internal-consistency

Normalization

Normalization skills enforce author-declared style and terminology. They emit a structured report plus a list of suggested edits. They do not auto-apply edits to the manuscript file.

Skill	One line	Primary grounding
terminology-normalization	Detect terminology drift across a manuscript and enforce the preferred / forbidden / synonyms lists declared in MANUSCRIPT_STATE.yaml.	internal-consistency, style-guides

Transformation

Transformation skills modify prose. Both shipped transformation skills inherit the same preservation contract — citations, statistics, declared terminology, declared core claims, and the author’s hedging stance are preserved or surfaced as per-edit notes. Both are explicit-invocation only and operate on a single named section at a time. The author reviews each diff and decides.

Skill	One line	Primary grounding
argumentative-flow	Improve a manuscript section’s logical and argumentative coherence while preserving every citation, statistic, and declared terminology choice.	reader-expectation-approach, narrative-frameworks, argument-mapping
compression	Propose page-limit-driven length reductions on a manuscript section while preserving every citation, every declared statistic, every declared core claim, and every declared terminology choice.	narrative-frameworks, semantic-preservation, copyediting-vs-developmental

Meta

Meta skills orient new users and explain scriptorium itself. They read no manuscript content and modify nothing.

Skill	One line	Primary grounding
explain	Explain scriptorium itself, a named skill, or a named MANUSCRIPT_STATE field.	—
tour	Conversational onboarding for new users.	—

Utility

Utility skills set up scriptorium state. They modify MANUSCRIPT_STATE.yaml (and only that file) when the author confirms.

Skill	One line	Primary grounding
init	Conversationally bootstrap MANUSCRIPT_STATE.yaml for a new or existing manuscript directory.	common-critiques-taxonomy

Lifecycle fit, summarised

Skills declare which document_phase values they operate on in their manifest.yaml#lifecycle_phases. They refuse cleanly on phases for which they do not have enough declared state to ground against. The phases listed under each row below are cumulative — skills invocable at draft are also invocable at review, revision, and submission (unless the skill explicitly narrows to a later phase).

Phase	Skills invocable
outline	explain, init, tour
draft	argumentative-flow, citation-audit, explain, figure-text-alignment, gap-finder, init, outlier-sentence-detector, reporting-guideline-fit, terminology-normalization, tour
review	argumentative-flow, author-contribution-audit, citation-audit, explain, figure-text-alignment, gap-finder, init, outlier-sentence-detector, reporting-guideline-compliance, reporting-guideline-fit, reviewer-simulation, terminology-normalization, tour, venue-fit
revision	argumentative-flow, author-contribution-audit, citation-audit, compression, desk-rejection-risk, explain, figure-text-alignment, gap-finder, init, outlier-sentence-detector, reporting-guideline-compliance, reporting-guideline-fit, reviewer-simulation, terminology-normalization, tour, venue-fit
submission	author-contribution-audit, citation-audit, compression, desk-rejection-risk, explain, init, outlier-sentence-detector, reporting-guideline-compliance, reviewer-simulation, tour, venue-fit

MANUSCRIPT_STATE.yaml#document_phase is set by scriptorium:init and is what the skills read.

Author-side only

Skills whose manifest.yaml declares positioning: author-side-only refuse to run on a manuscript the user did not author. Editorial-side use violates ICMJE, NIH, and major-publisher peer-review policy.

• desk-rejection-risk
• reviewer-simulation

Source of truth

This page is generated at docs-build time by docs/scripts/preprocess.py from every skills/<name>/manifest.yaml. To change what appears here, edit the manifest. Editing this file directly is wasted work — the next preprocess pass will overwrite it.