prose-audit

Diagnose semantic repetition, filler, and structural bloat in existing text.

14|1|Updated Jun 14, 2026
One-click install
npx skills add https://github.com/allemaar/open-skills --skill prose-audit-allemaar
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: prose-audit
Source: https://github.com/allemaar/open-skills/tree/main/skills/prose-audit
Command: npx skills add https://github.com/allemaar/open-skills --skill prose-audit-allemaar

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Text often consumes reader attention without adding meaning, evidence, or action. This Skill audits existing prose to identify semantic repetition, filler phrases, fake structure, stock AI rhetoric, and avoidable reading load, producing a severity-ranked diagnostic report without rewriting the text. ## Core Features & Use Cases - Pattern-Based Diagnosis: Classifies findings against a catalog of 258 stable pattern IDs (P001-P258) covering answer delay, manufactured structure, inflated diction, empty balance, and formatting load. - Severity Ranking: Labels each finding High, Medium, or Low based on whether it hides the verdict, inflates sentences, or adds stock texture. - Structured Report Output: Produces a verdict-first report with quoted examples, compression opportunities, justified density, and unchecked material. - Use Case: Paste a long AI-generated report and ask for an audit; receive a verdict plus the specific patterns (e.g., restatement loops, decorative tables, hedge stacking) that account for most of the bloat, then hand off to a rewrite skill. ## Quick Start Audit this document for AI filler and repetition, and report the highest-severity patterns with examples.

Frequently Asked Questions about prose-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find AI filler and bloat in my writing?

Run a prose audit that classifies each sentence by the job it performs, such as verdict, evidence, or action. Elements performing no job are flagged as removable load, grouped by pattern, and ranked by severity with quoted examples.

What patterns does a prose audit detect in text?

It detects 258 cataloged patterns including throat-clearing, restatement loops, fake taxonomies, nominalization, hedge stacking, empty balance, LinkedIn cadence, and decorative formatting. Each finding maps to a stable ID from P001 through P258.

Does the prose audit rewrite my text automatically?

No, the audit is diagnosis-only by design. It reports what adds no value and why, then recommends a separate rewrite step such as the human-rewrite skill to remove the diagnosed load while preserving every claim.

Can a prose audit prove text was AI-generated?

No, and it explicitly avoids authorship claims. One pattern is weak evidence; dense repeated clusters suggest a writing habit but never prove origin. The report covers only observed patterns, not attribution.

When should I not use a prose audit?

Skip it when you need rewriting rather than diagnosis, or when the text is already minimal. Necessary uncertainty, deliberate tone, and navigation in genuinely long documents are treated as justified payload, not defects.