Add methods reproducibility redline assistant

methods-reproducibility-redline/README.md

@@ -0,0 +1,28 @@
+# Methods Reproducibility Redline
+
+This module adds a focused AI-Assisted Research Tools slice for issue #13. It is a deterministic, synthetic-data-only assistant that turns a draft manuscript into a methods reproducibility packet before human peer review.
+
+The module covers:
+
+- paper summaries in abstract, executive, and layperson modes
+- domain inference for clinical, computational, and wet-lab drafts
+- peer-review diagnostics for missing method evidence
+- statistical, compliance, data, code, environment, and reagent redlines
+- citation recommendations with APA, MLA, and Nature-style insertions
+- reviewer tasks and audit digests for institutional review packets
+
+This is not another broad summarizer or generic citation formatter. The slice focuses on the handoff where an AI assistant must prove that the methods section is reproducible enough for review and must show exactly which evidence or citation is missing.
+
+## Local Validation
+
+```sh
+node methods-reproducibility-redline/test.js
+node methods-reproducibility-redline/demo.js
+```
+
+## Demo Evidence
+
+- [demo.mp4](demo.mp4) shows the problem, implementation scope, output packet, and validation commands.
+- [demo.svg](demo.svg) provides a static reviewer dashboard preview.
+- [requirements-map.md](requirements-map.md) maps the implementation to issue #13.
+- [acceptance-notes.md](acceptance-notes.md) lists reviewer checks.

methods-reproducibility-redline/acceptance-notes.md

@@ -0,0 +1,12 @@
+# Acceptance Notes
+
+Reviewer checks:
+
+1. Run `node methods-reproducibility-redline/test.js`.
+2. Run `node methods-reproducibility-redline/demo.js`.
+3. Confirm complete clinical drafts pass without blockers.
+4. Confirm risky computational drafts produce method, code, environment, and uncertainty redlines.
+5. Confirm citation recommendations include formatted references and insertion hints.
+6. Confirm `auditDigest` is stable for the same input.
+
+The implementation is dependency-free and uses synthetic manuscript examples only, so it can be reviewed without accounts, external corpora, or AI service keys.

methods-reproducibility-redline/demo.js

@@ -0,0 +1,30 @@
+"use strict";
+
+const { evaluateMethodReadiness } = require("./index");
+
+const draft = {
+  title: "Containerized literature triage for rare disease teams",
+  abstract:
+    "We evaluate a computational triage model that ranks rare disease papers for curator review.",
+  methods:
+    "The source code is available in a GitHub repository. The dataset version is RareLit snapshot 2026.04. The Python 3.12 runtime is captured in a Docker image. Statistical analysis reports bootstrap confidence intervals and limitations include English-language indexing bias.",
+  results:
+    "The model reduced curator screening load by 22% while preserving recall for known benchmark papers.",
+  keyFinding: "A reproducible triage pipeline can reduce curator review load without hiding evidence gaps.",
+};
+
+const result = evaluateMethodReadiness(draft, { citationStyle: "apa" });
+
+console.log("Methods reproducibility redline demo");
+console.log(JSON.stringify(
+  {
+    domain: result.domain,
+    readinessScore: result.readinessScore,
+    readyForPreReview: result.readyForPreReview,
+    redlines: result.peerReviewDiagnostics.redlines,
+    citationTopics: result.citationRecommendations.map((citation) => citation.topic),
+    auditDigest: result.auditDigest,
+  },
+  null,
+  2,
+));