Draft To Latex Block

Inputs

Required inputs:

• paper_blueprint (required): Verified paper-level question, answer, storyline, contributions, and target sections.
• contribution_list (required): Stable list of contributions already accepted by the authors as in-scope and evidence-backed.
• claim_evidence_map (required): Claim-to-evidence alignment table used to prevent unsupported statements and novelty inflation.
• source_text (required): Concrete local draft span to convert into LaTeX form without changing factual content.

Optional inputs:

• terminology_registry (optional): Canonical terms, symbols, and disallowed variants for consistency control.
• current_draft_context (optional): Neighboring draft text or outline used to determine heading level, notation reuse, and formatting consistency.
• venue_constraints (optional): Page limit, formatting rules, author preferences, or local template requirements.
• block_type_hint (optional): Intended LaTeX target such as section, subsection, paragraph, equation, table, figure, appendix_section, or auto.
• latex_context (optional): Existing preamble macros, equation conventions, float style, or table/figure formatting rules already used by the draft.

Output Contract

• Output format: block
• Required sections: latex_block, conversion_notes, preserved_claims, handoff_notes
• Tone and language: reviewer-friendly / English
• Length guidance: Keep outputs compact, directly reusable, and safe to paste into a larger .tex manuscript.

Workflow

1. Treat the provided inputs as the only source of truth.
2. Produce only the bounded artifact for this skill, not a whole-paper rewrite.
3. Preserve contribution names, terminology, and claim boundaries from the shared inputs.
4. Infer the local LaTeX block shape from block_type_hint when given; otherwise infer the smallest safe block from the source text.
5. Map structure conservatively:
   • headings -> \section{} / \subsection{}
   • ordinary prose -> plain LaTeX paragraphs
   • inline math -> $...$
   • display math -> \[ ... \]
   • tables -> table + tabularx or another width-aware layout with breakable columns
   • figure placeholders -> figure
   • appendix fragments -> \appendix followed by section-level content only when the source text actually starts an appendix boundary
6. Apply latex-layout defaults that reduce common formatting failures:
   • long identifiers or run names should not remain as raw unbreakable \texttt{...} spans; prefer a breakable wrapper such as a local \code{} macro built on \seqsplit, \path, or another line-break-safe representation
   • dense tables should prefer tabularx, p{}, or X columns over many rigid centered columns
   • table and figure placeholders should avoid overly tall boxed content that is likely to collide with nearby text
   • appendix-local floats should remain within the appendix boundary and should not drift past the references block
7. Run a local latex-layout sanity pass before finalizing:
   • check for likely overfull-line risks caused by long code-like tokens
   • check whether a table block is too dense for a single-column width and should be compacted or widened
   • check whether any appendix figure placeholder needs stronger float control or a nearby float barrier note
8. If you need the full original schema or prompt template, read references/spec.yaml.

Guardrails

Core hard rules:

• Do not invent experiments, results, contributions, baselines, theoretical conclusions, or citations.
• Maintain factual faithfulness, claim-evidence alignment, terminology consistency, overclaim suppression, and reviewer-friendly clarity.
• If evidence is missing, weak, or indirect, say so explicitly and reduce claim strength rather than filling gaps.
• Prefer bounded, reviewer-safe wording over broad or triumphalist phrasing.

Section-specific checks:

• Output stays within the scope of Style And Language Utilities and does not leak unsupported content from other sections.
• Contribution names, task setting, and method references match the global blueprint.
• Claims are phrased at the strength justified by the supplied evidence.
• Conversion must preserve scientific content while changing only local representation into LaTeX form.
• The converter must not silently upgrade placeholder assets into final figures, final tables, or final citations.
• Appendix conversion should only emit \appendix when the provided local block truly begins the appendix transition; otherwise emit only the appendix section body.
• Long identifiers, run names, config names, and script paths must be rendered in a breakable way; avoid emitting raw unbreakable \texttt{...} when line wrapping is likely to fail.
• Tables must prefer wrap-friendly columns and modest density; if a table is visibly too wide or too crowded, restructure it rather than squeezing more rigid columns into the block.
• Figure placeholders must not be so tall or float-unsafe that they are likely to collide with surrounding text or drift into later structural regions such as references.

Revision policy:

• Missing required input: Return a gap-marked scaffold and list exactly which missing inputs block a stronger output.
• Unsupported claim: Downgrade or remove the claim, explain the evidence gap, and never patch the gap with invented detail.
• Multiple valid options: Provide 2-3 bounded options with tradeoffs, but keep all options faithful to the same verified facts.

Coordination

Reasoning focus:

• logical coherence
• evidence grounding
• notation preservation
• latex safety
• editing efficiency

Self-checklist:

• Every main statement is supported by provided facts or explicitly labeled as a gap.
• No wording overstates novelty, empirical coverage, causality, or theory scope.
• Terminology, symbols, and contribution labels match the shared registry and blueprint.
• The emitted LaTeX block preserves the original claim set and does not introduce new content.
• Floats, math, and appendix transitions are only emitted when justified by the supplied source text.
• Long code-like tokens are breakable and unlikely to trigger line-end overflow.
• Table blocks use wrap-friendly columns and reasonable density for the target width.
• Appendix-local figure blocks are unlikely to drift past a bibliography or other end matter.

Upstream skills:

• 00_01_paper_blueprint_skill
• 00_04_claim_evidence_alignment_skill
• 00_06_terminology_notation_registry_skill
• 13_09_latex_friendly_sentence_skill

Downstream handoff:

• 14_07_full_draft_to_latex_pipeline_skill