paper-argument-reconstructor

Paper Argument Reconstructor

Scope

Use this skill when the user asks to improve a manuscript's logic, storyline, contribution framing, section structure, paragraph function, or experiment-to-claim alignment.

For full papers, major sections, introduction rewrites, contribution statements, or experiment-logic redesign, read references/argument_reconstruction.md. For Chinese-facing workflow guidance, references/argument_reconstruction_zh.md is available as a companion reference.

Use another skill when the task is primarily:

• TeX-safe expression polishing, proofreading, translation, or preservation checking: use paper-polisher (if available);
• idea novelty risk, dangerous baselines, reviewer attacks, or whether a contribution is defensible before manuscript writing: use idea-novelty-auditor (if available);
• literature review, related work, introduction background, research-gap synthesis, reference selection, citation-role assignment, or direct-competitor coverage: use problem-driven-literature-review (if available).
• audit-only checks of experiment-set minimality, ablation sufficiency, feasibility against stated resources, or report-style results narration: use experiment-section-auditor (if available).

If the user asks for novelty audit, literature review, and manuscript reconstruction together, use idea-novelty-auditor (if available) first, then problem-driven-literature-review (if available), then reconstruct only the defensible claims.

Verification Mode

This skill has two modes. By default it reconstructs as described above. When the input is an argument or section another tool or skill has already drafted or rewritten, switch to verification mode instead of regenerating:

• do not overwrite the other tool's wording;
• check it against this skill's rules: claim boundaries, contribution-to-evidence alignment, and whether any claim was strengthened beyond what the source supports;
• report only the deviations and their locations; fix one in place only if it breaks a hard rule.