writing-voice
Guide sentence-level expression while drafting or revising substantial prose so it remains recognizable as the writer: their vocabulary, humor, certainty, emotional emphasis, and allocation of attention. Use for prose that feels generic, corporate, AI-shaped, over-sanitized, or that no longer sounds like me; for tone and ear passes; and for local wording work in UI copy, documentation, or messages. Do not use to decide a document's content, structure, or destination.
Works with
--- name: writing-voice description: Guide sentence-level expression while drafting or revising substantial prose so it remains recognizable as the writer: their vocabulary, humor, certainty, emotional emphasis, and allocation of attention. Use for prose that feels generic, corporate, AI-shaped, over-sanitized, or that no longer sounds like me; for tone and ear passes; and for local wording work in UI copy, documentation, or messages. Do not use to decide a document's content, structure, or destination. license: MIT --- # Writing Voice ## Product Promise Make the writer easier to hear, not easier to mistake for a good generic writer. Voice is not polish applied after meaning and structure. It is evidence of what the writer means: their relation to the reader, their certainty, their humor, and what they linger on. The destination skill owns what belongs and in what order. This skill owns how that chosen material is expressed. ## Preserve The Grain Do not add, remove, reorder, promote, or demote material as a voice choice. Return those decisions to the destination. Different destinations have different native movements, so do not impose one house style on all of them. Keep the writer's concrete vocabulary, unusual phrasing, proportion of caveat to claim, jokes, fragments, repetitions, and detours when they do expressive work. A phrase may be grammatical but wrong if it changes the writer from curious to authoritative, conversational to formal, or amused to earnest. Preserving a few distinctive words is not enough when the draft has changed its posture. Repair roughness when it is merely in the way: a transcription error, a false start with no remaining work to do, or spoken order that makes the thought unintelligible. Do not repair something merely because smoother prose is available. Every substantial edit needs a positive account of the expressive job it performs, such as making a turn legible without removing its uncertainty. “This sounds better” is not an account. ## Draft, Then Rehear A first provisional draft may use an earned reveal, contrast, metaphor, or closing turn when it makes the supplied material easier to hear. It must not invent an event, fact, memory, reaction, certainty, or intention. Do not turn a sparse source into a polished generality. After a whole draft exists, read it at speaking speed. Mark generic connective tissue, flattened rhythm, and places where the attention moves away from what the writer cared about. Remove unsupported polish rather than merely improving it. Keep a vivid turn when the source earns it. When a reaction reveals a deeper mismatch, reread the complete source and the whole draft under that correction. Do not reduce “too solemn,” “too sanitized,” or “that is not why I said it” to a line edit. When a selected local revision is truly consequential but does not change the work's model, show the current and proposed wording, state the job it performs, and wait for the writer's decision. ## Match The Destination In UI copy, documentation, and messages, preserve the person while respecting the form; clarity never requires corporate scaffolding or false authority. Use punctuation to make movement legible. Prefer periods. Use colons for a real explanation, semicolons for closely related independent clauses, and parentheses for genuine asides. Avoid en dashes; default away from em dashes except for a real interruption or high-emphasis aside. The test is: did this expression make the writer easier to hear without normalizing the particular thing they were doing?
More Writing & Documentation skills
paper-context-resolver
lllllllama/rigorpilot-skills
Rigor Paper Context helper for README-first deep learning repo reproduction. Use only when the README and repository files leave a narrow reproduction-critical gap and the task is to resolve a specific paper detail such as dataset split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumption from primary paper sources while recording conflicts. Do not use for general paper summary, repo scanning, environment setup, command execution, title-only paper lookup, or replacing README guidance by default.
repo-intake-and-plan
lllllllama/rigorpilot-skills
Rigor Intake helper for README-first deep learning repo reproduction. Use when the task is specifically to scan a repository, read the README and common project files, extract documented commands, classify inference, evaluation, and training candidates, and return the smallest trustworthy reproduction plan to the main orchestrator. Do not use for environment setup, asset download, command execution, final reporting, paper lookup, or end-to-end orchestration.
minimal-run-and-audit
lllllllama/rigorpilot-skills
Rigor Run skill for README-first deep learning repo reproduction. Use when the task is specifically to capture or normalize evidence from the selected smoke test or documented inference or evaluation command and write standardized `repro_outputs/` files, including patch notes when repository files changed. Do not use for training execution, initial repo intake, generic environment setup, paper lookup, target selection, hidden scientific-meaning changes, or end-to-end orchestration by itself.

