| name | persuasion-audit-engine |
| description | Reviews persuasive writing across emails, DMs, LinkedIn posts, articles, landing pages, website copy, and high-stakes personal or professional messages. Use when the goal is to improve trust, clarity, response rate, conversion, emotional accuracy, or persuasive strength before rewriting. |
Persuasion Audit Engine
Use this skill to analyze writing before rewriting.
Apply it to:
- emails
- DMs
- LinkedIn posts
- LinkedIn articles
- landing pages
- website copy
- outreach messages
- founder messaging
- personal high-stakes communication
Default behavior:
- analyze first
- annotate second
- prioritize fixes third
- rewrite only if asked
Core principle
Persuasion is not manipulation.
Optimize for:
- trust
- clarity
- empathy
- relevance
- mutual benefit
- action
- relationship strength
- ethical influence
Do not optimize for:
- deception
- fake urgency
- fake scarcity
- emotional coercion
- invented proof
- pressure that damages trust
Inputs to identify or infer
Before reviewing, identify or infer:
- content type
- audience
- primary goal
- stakes
- voice requirement
If any are missing, infer them briefly from context.
Mode selection
Use one of two modes.
Fast pass
Use for shorter or lower-stakes writing, or when the user wants a quick review.
Return:
- Goal Read
- What’s Working
- What’s Weak, Missing, or Working Against the Goal
- Annotated Suggestions
- Top Fixes
- Rewrite Direction
Keep annotated suggestions brief and high-leverage.
Focus on the 3 to 5 most important lines, sections, or phrasing choices.
Deep dive
Use for longer, higher-stakes, or strategically important writing, or when the user asks for detailed analysis.
Return:
- Goal Read
- Audience Read
- Scorecard
- What’s Working
- What’s Weak, Missing, or Working Against the Goal
- Annotated Suggestions
- Top Fixes
- Rewrite Direction
Make annotated suggestions more comprehensive.
Cover the key lines, paragraphs, and structural choices affecting persuasion.
Default to fast pass unless:
- the user asks for depth
- the stakes are high
- the content is long
- the message would benefit from broader or line-by-line analysis
Review workflow
1. Goal Read
State the draft’s likely intended overall goals.
Include:
- the explicit goal
- any implied goal
- the action, response, belief shift, or emotional outcome the draft appears to want
Do not stop at the obvious ask.
Also identify subtext, relationship goals, trust goals, positioning goals, and unstated persuasion goals when relevant.
2. Audience Read
State what the audience likely:
- wants
- fears
- doubts
- resists
- needs to believe before acting
Use this in deep dive mode. In fast pass, include it only when needed to explain the critique.
3. Scorecard
Score each from 1 to 5 with one short reason:
- clarity
- relevance
- trust
- emotional resonance
- specificity
- credibility and proof
- objection handling
- differentiation
- friction reduction
- strength of ask
- channel fit
- overall persuasion strength
See reference/scoring.md for scoring guidance.
4. What’s Working
List the persuasive strengths already present.
5. What’s Weak, Missing, or Working Against the Goal
Always scan for:
- unclear ask
- weak stakes
- buried lead
- too much self-focus
- low audience empathy
- vague benefit
- weak proof
- weak differentiation
- no reason to act now
- friction in the next step
- no objection handling
- generic wording
- weak trust cues
- too much explanation
- not enough explanation
- tonal mismatch
- subtext that creates resistance
6. Persuasion lens check
Use the most relevant lenses for the context, not all equally.
Primary lenses:
- Cialdini, influence levers like reciprocity, authority, social proof, liking, commitment, scarcity
- Voss, tactical empathy, objection handling, calibrated language, emotional precision
- Carnegie, warmth, respect, audience-centered framing, dignity, relationship preservation
- Heath, clarity, simplicity, concreteness, memorability, stickiness
- broader persuasion and trust framing, audience state, resistance, context, outcome-based persuasion
For channel-specific priorities, see reference/channels.md.
Do not run a full hypnotic-writing pass unless the user asks for it or the content is clearly sales or marketing copy. Otherwise, only note major hook, curiosity, or momentum gaps briefly. For a full pass, see reference/hypnotic-pull.md.
7. Annotated Suggestions
Give keyed notes tied to exact lines, paragraphs, or sections.
For each note include:
- quoted line or section
- what it is doing now
- why it is weak, limited, or risky
- how to improve it
In fast pass, keep this to the highest-leverage 3 to 5 notes.
8. Top Fixes
Give the top 3 to 5 highest-leverage changes.
For each:
- issue
- why it matters
- likely impact if fixed
9. Rewrite Direction
Close with short guidance on what the next draft should do differently.
See reference/workflows.md for quick usage patterns and mode examples.
Output format
Use the mode that fits the context.
Fast pass:
- Goal Read
- What’s Working
- What’s Weak, Missing, or Working Against the Goal
- Annotated Suggestions
- Top Fixes
- Rewrite Direction
Deep dive:
- Goal Read
- Audience Read
- Scorecard
- What’s Working
- What’s Weak, Missing, or Working Against the Goal
- Annotated Suggestions
- Top Fixes
- Rewrite Direction
Guardrails
Do not:
- reward manipulation
- recommend fake scarcity
- invent proof
- flatten voice into generic sales language
- push pressure tactics that erode trust
Do:
- preserve voice by default
- note tradeoffs between persuasion and authenticity
- distinguish conversion writing from relationship writing
- prefer precision over intensity
- call out when softer language would work better
- call out when stronger language is needed
See reference/guardrails.md for the full ethics and calibration rules.
Examples and templates
For example reviews, see:
examples/email-example.mdexamples/linkedin-example.mdexamples/landing-page-example.md
For reusable invocation patterns, see:
reference/prompt-templates.md
