A modular AI-ready persuasion review skill that assesses, diagnoses, and improves persuasive writing before rewriting, grounded in principles from Cialdini, Voss, Carnegie, Heath, and Levine.
Persuasion Audit Engine analyzes writing for persuasive strength across:
- emails
- DMs
- LinkedIn posts and articles
- landing pages
- website copy
- founder and outreach messaging
- high-stakes personal messages
It is designed to:
- identify explicit and implied goals
- diagnose what is helping or hurting those goals
- return annotated suggestions
- prioritize the highest-leverage fixes
- preserve voice unless told otherwise
- fast pass by default
- deep dive for higher-stakes or longer writing
- no full hypnotic-writing pass unless requested or clearly relevant
- rewrite only if asked
The root SKILL.md acts as the orchestration layer. It stays concise and points to modular reference files only when they are needed.
persuasion-audit-engine/
├── SKILL.md # overview and operating rules – primary entrypoint
├── reference/
│ ├── channels.md # channel-specific review priorities
│ ├── scoring.md # scorecard guidance
│ ├── hypnotic-pull.md # optional stronger hook and momentum pass
│ ├── workflows.md # fast pass and deep dive usage patterns
│ ├── guardrails.md # ethics and calibration rules
│ └── prompt-templates.md # reusable invocation templates
├── examples/ # sample review formats
├── TEST-PROMPTS.md # test inputs for evaluation
├── SOURCES.md # framework influences and synthesis notes
├── CREATOR.md # creator attribution and package context
├── CHANGELOG.md # version history
└── PACKAGE-DESCRIPTION.md # package framing for sharing and registries
The repository root follows the Agent Skills specification.
Claude Code:
ln -s /absolute/path/persuasion-audit-engine ~/.claude/skills/persuasion-audit-engine
OpenAI Codex:
ln -s /absolute/path/persuasion-audit-engine ~/.codex/skills/persuasion-audit-engine
Then ask:
Use $persuasion-audit-engine to review this draft for persuasive strength before I rewrite it.
This skill keeps goal-awareness central without flattening the deeper persuasion read. It is built to preserve subtext, implied motives, tone leaks, trust signals, friction, and emotional dynamics.
Created by Jack Dalrymple, Founder of Cap & Cut. See CREATOR.md.
Released under the MIT License. See LICENSE.