Craft Forceful, Timeless English Prose
Style contract for forceful, timeless English prose - bans em dashes, filler transitions, and reversal-pivot cliches.
Why it matters
Generate high-quality, disciplined English prose that avoids modern AI tics and generic cadence. Ideal for essays, literary writing, and sharp rewrites requiring a concrete, confident voice.
Outcomes
What it gets done
Produce clean, exact, and muscular prose.
Rewrite text with strong style discipline.
Apply specific stylistic registers (e.g., literary_modern, cold_steel).
Ensure output is readable at speed and rewarding on reread.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/ag-beautiful-prose | bash Overview
Beautiful Prose (Claude Skill)
A style contract banning em dashes, reversal-pivot cliches, filler transitions, therapy language, and AI tells, with four voice registers and a manual lint checklist for forceful prose. Use for essays, literary writing, or sharp rewrites needing real stylistic discipline, not for casual replies or technical docs.
What it does
A style contract for producing forceful, timeless English prose free of modern AI tics, treated as a hard pass/fail specification, not a vibe. Activated by prepending "Apply the Beautiful Prose skill" to a request, with no acknowledgment, just the prose, optionally tuned via four control tags: REGISTER (founding_fathers, literary_modern, cold_steel, or journalistic, defaulting to literary_modern), DENSITY (lean, standard, or dense), HEAT (cool, warm, or hot), and LENGTH (micro, short, medium, or long). It bans six categories outright: em dashes rendered as a double hyphen (use periods, commas, colons, semicolons, or line breaks instead); "it's not X, it's Y" reversal-pivot constructions and their masked variants such as "not X but Y" or "X is a symptom, Y is the cause" used as a cheap reversal; filler transitions and scene-setting phrases like "at its core," "in today's world," "that said," "let's explore," and "ultimately"; therapeutic or validating language such as "I hear you," "you're valid," and "be kind to yourself"; AI tells and meta-commentary like "in this essay," "this piece explores," "here are the key takeaways," and style apologies; and symmetry padding, meaning balanced sentences for their own sake and decorative three-part lists. Positive constraints push declarative sentences with aggressively varied length, concrete nouns over abstractions, strong verbs over adverbs, Anglo-Saxon word weight with Latinate precision only when it buys accuracy, paragraphs that open with substance rather than a hook and close without restating the thesis, and an authoritative voice that avoids hedging and doesn't moralize. Four named registers each carry a distinct texture: founding_fathers (formal, spare, civic gravity, balanced but undecorated syntax), literary_modern (vivid lean imagery, controlled heat, minimal ornament), cold_steel (severe compression, punchy, unsentimental, high signal low warmth), and journalistic (crisp, factual, clean narrative momentum, no clickbait cadence). A pre-finalization quality check strips template-assembled lines, redundant restatements, emotion-steering sentences, and any paragraph that doesn't advance meaning, explicitly preferring silence to slop when quality is uncertain. A manual lint checklist fails output containing a literal em-dash substitute, a reversal-pivot pattern, a banned filler transition, therapy language, meta writing-talk, or five consecutive sentences of similar length. Worked before-and-after examples contrast the banned reversal pivot against its fix, and banned filler against its fix.
When to use - and when NOT to
Use for prose or rewrites that need strong style discipline and no generic AI cadence - essays, literary-style writing, sharp rewrites, or exacting English prose where a forceful, concrete voice is wanted instead of friendly assistant-style copy. Not for casual conversational replies, technical documentation, or any context where the banned constructions such as transitions, hedging, or bulleted structure are actually useful rather than filler - the skill's own output rules default to plain prose with no headings or bullets unless explicitly requested.
Inputs and outputs
Input is the activation phrase plus optional REGISTER, DENSITY, HEAT, and LENGTH control tags and the actual writing request. Output is plain-text prose, no headings or bullets by default, that passes the manual lint checklist: no em dashes, no reversal pivots, no filler transitions, no therapy language, no meta commentary, no mechanically balanced sentence lengths. Test cases live in references/test-cases.md.
Integrations
None - this is a self-contained prompting and style contract with no external tool dependencies, referencing only its own companion test-cases file.
Who it's for
Writers and editors using an AI assistant who want prose with real stylistic teeth - essays, sharp rewrites, literary copy - and a concrete, checkable list of banned constructions rather than a vague "write better" instruction.
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.