feat(academic-aio): sections 10-12 + Phase 2/3 — schema markup, summary boxes, funder OA, audit tooling

[Yoojin-nam]

Summary

Two-commit operationalization of the academic-aio skill, expanding it from a single SKILL.md to a full reference + tooling bundle informed by a real KJR multimodal-LLM review post-mortem (64 citations / 7 months).

Commit 1 — Sections 10-12 + KJR case study + AIO_GENERAL checklist

• SKILL.md +130 lines: §10 Q&A and Entity-Extraction Optimization, §11 First-Mover Timing and Citation-Graph Density, §12 Cross-Platform Launch Sequencing.
• references/case_studies/kjr_mllm_2025.md — driver analysis, structural features extracted by AI search, lesson-to-rule mapping, replication checklist.
• references/checklists/AIO_GENERAL.md — YAML+Markdown machine-readable audit checklist covering all sections 1-12.

Commit 2 — Phase 2/3 references + tooling

• references/journal_summarybox_templates.yaml — Lancet DH / Radiology / Radiology AI / npj DM / Nature Medicine / JAMIA summary-box patterns with last_verified dates and USER MUST VERIFY guards.
• references/schema_markup_templates/ — JSON-LD for ScholarlyArticle, SoftwareSourceCode, Dataset, Person + embed README.
• references/oac_funding_checklist.yaml — Plan S / NIH / UKRI / Gates / Wellcome / NRF / MoHW open-access compliance matrix with 4-step decision tree.
• references/reporting_guideline_mapping.md — AIO rule x TRIPOD+AI / CLAIM 2024 / STARD-AI / TRIPOD-LLM / DECIDE-AI item-level crosswalk so /check-reporting and /academic-aio audits share evidence.
• scripts/validate_schema.py — JSON-LD validator CLI (placeholder-aware DOI/ORCID checks). Tested on all four templates: PASS.
• scripts/batch_metadata_audit.py — multi-repo + HF card audit (README structure, CITATION.cff, LICENSE, YAML keys, prose sections, PHI patterns). Smoke-tested.
• templates/aio_audit_checklist.md.j2 — Jinja2 audit-table template.
• SKILL.md cross-reference wiring at §1.6, §2.1, §3.4, §5.6, §9 to all new artifacts; §9 strengthens the /check-reporting invocation rule and adds explicit §11/§12 invocation at submission planning and post-acceptance.

Motivation

The skill previously documented AIO concepts thoroughly in SKILL.md but had no operational artifacts (checklists, templates, scripts). A real KJR review with unexpectedly high citation accrual let me identify which AIO levers actually mattered (OA + PMC dual indexing, taxonomy structure, explicit challenge statements, fast publication, multi-disciplinary roster) and codify them into:

1. New SKILL.md sections 10-12 with prescriptive guidance.
2. References folder so anti-hallucination guards (USER MUST VERIFY, last_verified) prevent journal/funder policy drift.
3. Scripts so audits are reproducible across multiple papers / repos / cards in a team setting.
4. Cross-skill wiring so /academic-aio and /check-reporting no longer duplicate work.

Test plan

• validate_skills.sh pre-commit hook PASS for both commits (academic-aio: 1 WARN on quality gates count, all other gates PASS)
• python scripts/validate_schema.py references/schema_markup_templates/*.jsonld — PASS on all four templates
• python scripts/batch_metadata_audit.py --hf-card README.md — smoke test, output well-formed JSON
• Dry-run /academic-aio invocation on a current manuscript draft (e.g., next CBCT meta-analysis or RFA-Adjunct submission) to verify §10-12 trigger correctly
• Manual review: ensure no PII / personal identifiers in case study (corresponding author and institution intentionally abstracted; only public paper title / DOI / PMC ID retained)
• Cross-skill compatibility: verify /check-reporting and /write-paper Phase 7 still pair cleanly with the expanded checklist (use reporting_guideline_mapping.md as the integration contract)

🤖 Generated with Claude Code