What problem does it solve? AI-assisted drafts of papers and grant proposals come out generic and verbose, with formulaic openers, inflated phrasing, over-claimed results, and drift from the author's own voice, while general-purpose humanizers flatten the precision that scholarly writing depends on. ## Core Features & Use Cases - Six-layer editing pass: audits and rewrites text across general AI tells, academic-specific tells (over-claiming verbs, significance hype, formulaic openers), scholarly-convention preservation, claim-to-evidence matching, voice/venue calibration, and a dedicated NSF/NIH proposal mode. - Claim-evidence discipline: downgrades verbs stronger than the data, attaches numbers, ranges, and citations to vague claims, and never alters any number, equation, or reference. - Funding-proposal mode: keeps the vision language a paper would trim, enforces claim-to-feasibility matching, and prioritizes the first pages reviewers score (NIH Specific Aims, NSF Project Summary). - Use Case: Paste an AI-drafted abstract or Specific Aims page, optionally supply a prior accepted paper for voice matching and a target venue or agency, and receive a cleaned rewrite plus a change log of patterns removed and claims softened. ## Quick Start Ask the agent to run the academic-humanizer skill on your draft section or main.tex, optionally specifying a prior paper for voice matching and the target venue or funding agency.