What problem does it solve? Research manuscripts in the FLARE, PFR and Voltivity projects repeatedly arrive with British spelling inherited from the cited literature, conversational sentence-style headings, and machine-generation tells such as invisible Unicode characters and formulaic phrasing. This Skill provides a systematic language and register pass that fixes these issues without breaking LaTeX macros, citations, or page layout. ## Core Features & Use Cases - American English conversion: Mask-aware British-to-American spelling conversion that protects siunitx macros like \metre, bibliographies, control sequences, math, and comments before substituting. - Heading normalization: Rewrites \section, \subsection and \paragraph headings as noun phrases, preserving labels such as G1: or D3: and moving claims into body text. - Machine-generation tell detection: Layered ripgrep checks for zero-width characters, typographic Unicode substitutes, phrasing tells, and AI provenance in commits and PDF metadata. - Prose shape analysis: Flags undeveloped paragraphs, clause stacks, uniform sentence length, and overlong blocks using measurable thresholds. - Post-edit verification: Proves \SI, \label, \cite and \ref multisets are identical, page count is unchanged, and no number moved after any prose edit. - Use Case: Before submitting a FLARE technical note, run a full language pass that converts 127 instances of fibre to fiber without touching 652 \metre macros, rewrites sentence-style headings, and verifies the build is byte-equivalent in structure. ## Quick Start Run a full prose-style pass on my manuscript.tex covering Americanization, headings, machine-generation tells, and prose shape, then verify the LaTeX invariants are unchanged.
