Source and execution ledger

Reviewed 2026-10-01. These are dated snapshots, not a promise that the next submission cycle has identical rules. This skill has no remote API client: its Python helpers only read local templates/PDFs and invoke Poppler. Publisher and conference links are human-facing documentation or template downloads, not submission endpoints.

Journal sources

Discovery hubs were reviewed for their role, not treated as uniform formatting rules: IEEE templates, BMC, Frontiers, PNAS, and APS.

Conference sources

Other families remain discovery links with no cached limits. The broken USENIX and ISMB generic URLs were replaced with USENIX conferences and ISMB series; EMNLP resolves through SIGDAT. AAAI, IJCAI, KDD, SIGIR, RECOMB, PSB, and IEEE conference/ICRA hubs were checked; select the actual year's call before drafting.

Grants and reporting

• NSF PAPPG and Chapter II: 24-1 remains the effective base, with 26-200 and 26-202. The Research.gov DMSP tool supersedes the old two-page PDF process from 2026-04-27. Common-form/SciENcv workflows replace obsolete biosketch layouts. Prior-support publication lists are not capped at five publications.
• NIH page limits and attachment formatting: generally one-page Aims, R01 Strategy 12 pages, R21 Strategy 6; body text at least 11 points, margins at least half an inch, density limits apply. Actual PDF and NOFO remain controlling.
• NIH Common Form and 26-079: implementation for 2026-01-25 due dates, system enforcement from 2026-05-08, research-security-training certification for 2026-05-25 due dates. These are separate dates.
• DOE Office of Science guidance and DARPA Heilmeier questions are starting points; the active office-specific call determines submission system and limits.
• CONSORT/SPIRIT: current core guidelines are 2025, not CONSORT 2010/SPIRIT 2013. STROBE, PRISMA, and STARD require design-appropriate checklists/extensions, not a universal medical-paper structure.

Tool and template execution

The maintained helpers require only Python's standard library. Poppler's pdffonts table includes all font rows and an emb flag; a listed name alone is not proof of embedding. Local pdfinfo/pdffonts execution checked both embedded-font PDFs and an intentionally nonembedded Helvetica fixture. The inspector still cannot certify point sizes, margins, reading order, anonymity, or the manually supplied content-page boundary.

Native pdfLaTeX (TeX Live 2025), with shell escape disabled, compiled all nine bundled .tex files. The NeurIPS wrapper used the actual 2026 author kit and a synthetic bibliography fixture; it is not a paper ready to submit. Elsevier used the current 3.5 class generated from upstream source. BibTeX separately exercised all three bundled styles. Plain elsarticle-num does not supply author names for \citet; the numeric-with-names and author-year styles do.

The Elsevier samples are unmodified upstream copies and repeat fig1 on a table and figure, producing a duplicate-label warning. Assign unique labels in the working manuscript. Their empty graphical-abstract/highlight placeholders are also examples, not completed content. Nonfatal epstopdf shell-escape warnings are expected with no external conversion enabled; PLOS's microtype/footnote patch warning and tikzposter's nullfont diagnostic were recorded during review.

beamerposter, tikzposter, and qrcode package documentation informed the poster workflow. The beamerposter/tikzposter preambles and local QR generation compiled. A separate baposter preamble compatibility check also compiled using a cached archival v2.0 class dated 2011-11-26 from the mloesch mirror; this is not evidence of a current maintained upstream release. The NIH Aims placeholder is one page; the poster is one A0 portrait page with a visible title. Actual authored content must be recompiled and visually checked. Poster layout/type-size advice is heuristic. WCAG contrast guidance is a web accessibility criterion, not certification that a printed poster or PDF is accessible. Color simulations and asset-library links are optional author resources, not tested integrations or permission to redistribute their artwork.

Verification boundaries

• Fresh Science instructions could not be retrieved. NEJM and Annals returned navigation without complete article-type instructions. Unverified fixed numerical limits were removed; retrieve current full instructions before submitting.
• The generic ACM/CHI hubs blocked automated access; official CHI and KDD 2026 instructions supplied the current acmart workflow evidence.
• baposter.cls is absent from the installed TeX distribution and its current package page could not be retrieved. The short preamble has only the archival compatibility check described above; no baposter class/template is bundled.
• Browser-based design services, submission portals, authenticated SciENcv/Research.gov workflows, print-shop output, and real submission acceptance were not exercised. Source verification and local compilation are different evidence.
• All prose examples contain placeholders. They neither report executed experiments nor establish scientific validity, journal fit, or acceptance.