seandavi
- 16 skills
- 0 followers
- 6 hours ago last updated
- ▌ Citation Audit · seandavi bundleAudit existing citations in a manuscript for claim-support alignment, primary-vs-review mismatch, causal overreach, and unsupported assertions. Reports findings as structured markdown. Does NOT add or invent citations.
- ▌ Argumentative Flow · seandavi bundleImprove a manuscript section's logical and argumentative coherence while preserving every citation, statistic, and declared terminology choice. Produces a structural diagnosis, a proposed outline, the revised text with a diff against the source, and a preservation report. Invoke ONLY when the user explicitly asks for argumentative-flow improvement on a specific section.
- ▌ Desk Rejection Risk · seandavi bundleAuthor-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 problems editors triage on. Outputs a structured markdown report with a qualitative risk band. NOT for editorial-side use — running this on someone else's manuscript violates ICMJE / NIH / Elsevier / Nature policy.
- ▌ Reviewer Simulation · seandavi bundleAuthor-side simulation of peer review across four attentional lenses (methodological skeptic, domain expert, translational/clinical, statistical). Surfaces likely major and minor critiques, fatal concerns, enthusiasm drivers, and concrete revision suggestions. Output is structured markdown. NOT for editorial-side use — running this on someone else's manuscript violates ICMJE / NIH / Elsevier / Nature policy.
- ▌ Figure Text Alignment · seandavi bundleAudit text-only alignment between figure captions and the body-text sentences that reference them. For every figure, classify the caption ↔ body-text-reference pair as aligned / partially aligned / misaligned / cannot determine, and flag pattern-level issues (orphan figure, phantom figure, panel mismatch, axis/units divergence, direction divergence). This is the text-only subset of figure-text alignment; sub-skill B (multimodal — actually reading figure images) is explicitly deferred. Does NOT read figure images. Does NOT modify the manuscript.
- ▌ Reporting Guideline Fit · seandavi bundleRead the manuscript's methods section (and abstract/title for context) and infer which EQUATOR Network reporting guideline applies — CONSORT for RCTs, STROBE for observational, PRISMA for systematic reviews, ARRIVE for animal research, TRIPOD/TRIPOD+AI for prediction models, STARD for diagnostic accuracy, CARE for case reports, COREQ for qualitative, with relevant AI-extensions (CONSORT-AI, SPIRIT-AI, TRIPOD+AI) when applicable. Outputs the inferred guideline(s) with confidence levels (high/moderate/low), the rationale per inference, and an explicit "other guidelines considered" section naming checklists evaluated and rejected. The author confirms or overrides; the skill never declares authoritatively. Upstream inference for the v0.3 reporting-guideline-compliance skill — this skill stops at "which checklist applies?"; running the full checklist against the manuscript is a separate skill. Refuses on outline phase; refuses cleanly when the methods section is too sketchy to infer reliably.
- ▌ Author Contribution Audit · seandavi bundleAudit the manuscript's Author Contributions section against ICMJE's four authorship criteria and CRediT's 14 contributor roles. Flag what's missing per author, suggest how to map who-did-what to canonical CRediT roles, and (when `target_venue` is set) compare against journal-specific variants. Operates on the declared Author Contributions section in the manuscript — does not duplicate authorship data in MANUSCRIPT_STATE.yaml, does not auto-write or rewrite the section. Outputs structured markdown with soft recommendations. Invoke when the author asks for a contributions check, is preparing for submission, or wants to verify the section meets a venue's requirements. Refuses on outline phase; refuses to adjudicate authorship disputes; refuses editorial-side use on someone else's manuscript.
- ▌ Outlier Sentence Detector · seandavi bundleFlag sentences that are statistical outliers in the manuscript along three narrow dimensions — length, syntactic complexity, and nominalization density — for human review. Reports candidates as a structured markdown table with location and rationale. Does NOT produce a quality score, a grade, or a pass/fail. Does NOT edit the manuscript. Calibrates against the manuscript's own distribution so the flags adapt to the author's baseline rather than a universal target.
- ▌ Terminology Normalization · seandavi bundleDetect terminology drift across a manuscript and enforce the preferred / forbidden / synonyms lists declared in MANUSCRIPT_STATE.yaml. Reports inconsistencies with exact occurrence locations and suggests normalizations. Does NOT modify the manuscript without explicit author consent.
- ▌ Reporting Guideline Compliance · seandavi bundleRun an EQUATOR Network reporting-guideline checklist (CONSORT, STROBE, PRISMA, ARRIVE, STARD, TRIPOD/TRIPOD+AI, CARE, COREQ, CHEERS, plus AI extensions) against a manuscript and classify every checklist item as present / partial / missing / not-applicable. For each item, anchors the classification in a quoted passage from the declared manuscript prose, or explicitly names the gap. Downstream of the v0.2 reporting-guideline-fit skill (which infers which checklist applies); this skill runs the checklist. Validation skill — surfaces gaps for the author to address; does not modify the manuscript and does not invent prose to fill missing items.
- ▌ Init · seandavi bundleConversationally bootstrap MANUSCRIPT_STATE.yaml for a new or existing manuscript directory. Infers what it can from the filesystem (title, sections, bibliography paths) and asks the author about the subjective fields (core_claims, known_weaknesses, terminology, audience). Validates against the JSON Schema before writing. Invoke when the user wants to set up scriptorium on a manuscript that does not yet have — or has only a partial — MANUSCRIPT_STATE.yaml.
- ▌ Tour · seandavi bundleConversational onboarding for new users. Three or four turns. Greets, asks what the user is working on (manuscript, grant, review, or exploring), walks through the relevant subset of scriptorium, shows an example MANUSCRIPT_STATE.yaml, and ends with a concrete next move (usually scriptorium:init). Does no work on the user's project — strictly read-only. Designed to be the single skill a maintainer can point a new user at instead of linking them to documentation. Invoke when the user says they're new to scriptorium, asks "what is this", runs the plugin for the first time, or whenever someone wants a guided walk-through.
- ▌ Explain · seandavi bundleExplain scriptorium itself, a named skill, or a named MANUSCRIPT_STATE field. With no argument, gives a whole-system overview (what scriptorium is, what each skill does, how the shared state ties them together). With an argument, drills into that skill or field — what it does, what it consumes, what it produces, and the grounding notes behind its design. Reads SKILL.md frontmatter and the referenced grounding files; no manuscript content is consumed. Use when the user wants to understand scriptorium before invoking it, or when a leaf skill has prompted them with "want a tour of <skill> before you run it?"
- ▌ Venue Fit · seandavi bundleAssess venue fit for a manuscript and return a tiered set of candidate venues (likely fit / stretch / probably premature) with per-venue rationale, fit caveats per axis (scope, audience, methodological, novelty, significance, open-access/cost), and explicit predatory-venue refusal. Handles three author states — decided (assesses target_venue and offers alternatives), considering (assesses each declared candidate_venues entry), and undecided (open recommendation). Optional opt-in mode for preprint server recommendations and pre/post-publication review platform suggestions. Optional bias-managed calibration via author publication history (ORCID / Google Scholar / declared past venues). Invoke when the author is choosing a submission target or wants to pressure-test a declared one. Refuses to recommend predatory venues; refuses to operate without `project.target_type`; refuses on outline-phase manuscripts.
- ▌ Gap Finder · seandavi bundleIdentify gaps in existing draft prose — claims under-supported, premises missing, related work not engaged with, counterarguments not addressed, internal-consistency mismatches — and suggest directions for filling them. Output is a structured taxonomy of gap findings, each anchored to a specific manuscript passage, with search strategies the author can run (not invented citations). Refuses cleanly when asked to operate on prose that doesn't exist yet (empty sections, "what should I write about" questions) — the author owns the proposer step. Invoke when the user explicitly asks "what's missing here?" / "what gaps are there?" / "what's the discussion not engaging with?" on a manuscript at draft phase or later.
- ▌ Compression · seandavi bundlePropose page-limit-driven length reductions on a manuscript section while preserving every citation, every declared statistic, every core claim, and every declared terminology choice. Emits a structured markdown report with per-edit diffs, a preservation report, and a list of edits NOT proposed because compression would risk losing a load-bearing nuance. Suggests edits; never auto-applies. Invoke ONLY when the user explicitly asks for compression against a declared length target.