codex-paper-skills

Calibrate academic writing style based on external samples. Extract complete style profile from reference, diagnose target paper, and generate restructured copy. Always performs full analysis (deep) and complete pipeline (profile → diagnosis → restructure). Outputs to ./paper-style/runs/<timestamp>-<target>/.

Apply a controlled human-style post-polish pass only after a formal full-paper polish has already been completed. Use when the user wants a finished academic paper to sound less mechanical by adding restrained natural phrasing in Introduction, Related Work or Related Works, and Results sections, with optional minor grammar imperfections only when the user explicitly enables grammar-error mode. Before editing, remind the user that paper-polish-workflow, paper-refine-special-en, or paper-refine-special-zh should be used first for formal polishing; preserve facts, citations, LaTeX syntax, formulas, and technical correctness.

Use when experiments complete to judge what claims the results support, what they don't, and what evidence is still missing. A paper architect/reviewer evaluates results against intended claims and routes to next action (pivot, supplement, or confirm). Use after experiments finish — before writing the paper or running ablations.

Compile LaTeX paper to PDF, fix errors, and verify output. Use when user says \"编译论文\", \"compile paper\", \"build PDF\", \"生成PDF\", or wants to compile LaTeX into a submission-ready PDF.

Generate publication-quality figures and tables from experiment results. Use when user says \"画图\", \"作图\", \"generate figures\", \"paper figures\", or needs plots for a paper.

Generate a structured paper outline from review conclusions and experiment results. Use when user says \"写大纲\", \"paper outline\", \"plan the paper\", \"论文规划\", or wants to create a paper plan before writing.

Review paper text or PDFs from a final-check or reviewer perspective. Use when the user wants fatal logic checks on English LaTeX, or a harsh reviewer-style report on a full paper PDF with concrete rejection risks and revision advice.

Draft paper content section by section from a paper plan, first as complete Markdown drafts and then as one LaTeX file per section. Use when user says \"写论文\", \"write paper\", \"draft LaTeX\", \"开始写\", wants Markdown-first paper drafting, or wants to generate LaTeX content from a problem-aligned paper plan.

Workflow 3: Full paper writing pipeline. Orchestrates paper-plan → paper-figure → paper-write Markdown drafts → paper-write LaTeX section files → paper-compile → paper-review → formal polish → optional human-style polish → proof/claim/citation audits to go from a narrative report to a polished PDF. Use for /goal-based long-running paper writing, "写论文全流程", "write paper pipeline", "从报告到PDF", "paper writing", Markdown-first paper drafting, or complete paper generation and revision. At `— effort: max | beast` (or explicit `— assurance: submission`), Phase 6 gates the Final Report on required audit artifacts and blocking verdicts.

Rigorous mathematical proof verification and fixing workflow. Reads a LaTeX proof, identifies gaps via AGENTS.md paper architect/reviewer audit, fixes each gap with full derivations, re-reviews, and generates an audit report. Use when user says "检查证明", "verify proof", "proof check", "审证明", "check this proof", or wants rigorous mathematical verification of a theory paper.

Review and improve a formal academic rebuttal letter draft for peer review responses.

Write a formal academic rebuttal letter from rebuttal strategy notes and reviewer comments.

Zero-context verification that every bibliographic entry in the paper is real, correctly attributed, and used in a context the cited paper actually supports. Uses the AGENTS.md paper architect/reviewer role with web/DBLP/arXiv lookup to catch hallucinated authors, wrong years, fabricated venues, version mismatches, and wrong-context citations (cite present but the cited paper does not establish the claim). Use when user says "审查引用", "check citations", "citation audit", "verify references", "引用核对", or before submission to ensure bibliography integrity.

Use when main results pass experiment-result-to-claim (`claim_supported = yes` or `partial`) and ablation studies are needed for paper submission. The paper architect/reviewer designs ablations from a reviewer's perspective; the paper executor reviews feasibility and implements.

Zero-context verification that every number, comparison, and scope claim in the paper matches raw result files. Uses a fresh paper architect/reviewer with no prior context to prevent confirmation bias. Use when user says "审查论文数据", "check paper claims", "verify numbers", "论文数字核对", or before submission to ensure paper-to-evidence fidelity.

Turn a stable research proposal or method idea into a detailed, claim-driven experiment roadmap. Use after `paper-plan`, or when the user asks for a detailed experiment plan, ablation matrix, evaluation protocol, run order, compute budget, or paper-ready validation that supports the core problem, novelty, simplicity, and any LLM / VLM / Diffusion / RL-based contribution.

Generate an image-generation prompt, show that prompt to the user, and then directly use the same prompt to generate a raster reference image for an academic algorithm architecture figure from paper notes or existing code. Use when the user wants a paper method/framework/architecture diagram reference, especially for deep learning papers where the default visual language should follow a clean Transformer-style architecture.

Advise on paper figures, charts, and captions. Use when the user wants experimental chart-type recommendations, figure captions, or table captions for academic papers.

Structures and derives research formulas when the user wants to 推导公式, build a theory line, organize assumptions, turn scattered equations into a coherent derivation, or rewrite theory notes into a paper-ready formula document. Use when the derivation target is not yet fully fixed, the main object still needs to be chosen, or the user needs a coherent derivation package rather than a finished theorem proof.

Workflow 4: Submission rebuttal pipeline. Parses external reviews, enforces coverage and grounding, drafts a safe text-only rebuttal under venue limits, and manages follow-up rounds. Use when user says "rebuttal", "reply to reviewers", "ICML rebuttal", "OpenReview response", or wants to answer external reviews safely.

Analyze experimental results for academic papers and give model-selection guidance. Use when the user wants LaTeX experiment analysis paragraphs from data, or practical guidance on which model family to use for academic writing, coding, or paper-polish workflows.

Check academic drafts against target-journal style and submission expectations. Use when the user mentions a target journal, asks for abstract or highlights formatting, wants title or keyword guidance, or needs cross-section consistency checks before submission.

Run an interactive academic-paper polishing workflow with explicit checkpoints. Use when the user asks to polish step by step, confirm structure before rewriting, revise sentence logic incrementally, compare multiple expression options, or do a gradual abstract or section refinement instead of an immediate full rewrite.

Refine academic writing at the paragraph level. Use when the user wants English LaTeX polishing, Chinese paper polishing, shortening, expansion, or de-AI rewriting for LaTeX or Word-ready text.

Perform high-intensity global polishing for English academic papers in LaTeX. Use when the user wants section-level or document-level rewriting with global logic checking, outline reconstruction, compression, sentence-level polishing, and a final reviewer-style pass, or when the user explicitly asks for a staged structure-to-logic-to-expression English polish workflow with confirmation checkpoints.

Perform high-intensity global polishing for Chinese academic writing in either Word-style plain text or Chinese LaTeX. Use when the user wants section-level or document-level rewriting with global logic checking, outline reconstruction, compression, sentence-level polishing, and a final reviewer-style pass, or when the user explicitly asks for a staged structure-to-logic-to-expression Chinese polish workflow with confirmation checkpoints.

Translate and rewrite academic paper text across Chinese and English. Use when the user wants Chinese-to-English LaTeX translation, English-to-Chinese literal translation, or Chinese academic paragraph rewriting for Word-style manuscripts.