MindFluence

[ English | Русский ]

Persuasion engineered through psychology, not manipulation.

An AI prompt-skill that turns 20 cognitive biases into high-converting marketing copy - for any language, any platform, any audience. v2.1: self-contained single-file with narrative depth. All tables, 13 anti-patterns with detection rules, cultural data, and few-shot examples are inlined into SKILL.md. Named Person Story Arc (NPSA) requirement ensures human-quality output alongside mechanical verification. No external file reads required for core operation.

---

What is this?

MindFluence is a system prompt (skill) for LLMs - GPT, Claude, Gemini, DeepSeek or any capable model. It transforms the AI into a marketing strategist who understands why humans click, read, and buy - not just what to write.

It's built on decades of research in behavioral economics and evolutionary psychology: Kahneman's two systems, Cialdini's principles, Munger's psychological misjudgments, Festinger's cognitive dissonance - distilled into actionable copywriting instructions.

How it works

1. Load SKILL.md as a system prompt into any LLM. That's it - the file is fully self-contained.
2. Give a task: "Write a LinkedIn post about..." / "Create a landing page for..." / "Audit this ad for cognitive biases."
3. Get output with every bias labeled, every decision explained, every word psychologically justified.

The LLM uses a single-table Bias Selection Router (audience × product × platform → instant bias stack) instead of a 10-step procedure. After generating, it runs a mechanical Post-Generation Verification - 6 checks (numbers, names, exit, explain, blame-system, human narrative) — and a mandatory Named Person Story Arc requirement before delivering.

Four modes:

• Quick Mode (default) - 3-step internal procedure: lookup router → scenario override → verify. Fast for user, fast for LLM.
• Deep Mode - the AI asks 4 clarifying questions at once, then routes through the same lookup.
• Audit Mode - analyzes existing copy for biases and anti-patterns using mechanical detection rules (13 anti-patterns including Statistical-Only Fallacy).
• Optimize Mode - iterates on underperforming copy using funnel-to-bias mapping. Generates A/B variants with statistical confidence thresholds.

The 20 Cognitive Biases

#	Bias	Category	Marketing Use
1	Social Proof / Bandwagon	Social	Used by teams at Google, Airbnb, and 10,000+ startups
2	Anchoring	Optimizer	Enterprise: $499/mo → Pro: $99/mo
3	Framing	Filter	"90% success rate" vs "Only 10% fail"
4	Appeal to Authority	Social	The same method taught at Harvard Business School
5	Fear / Loss Aversion	Filter + Social	Every day without X costs you $200 in missed revenue
6	Availability Heuristic	Optimizer	"Sarah tripled her revenue in 3 months. Here's the exact playbook."
7	Confirmation Bias	Filter	"You already know newsletters are broken. Here's the data that proves your instinct right."
8	Cognitive Dissonance	Filter + Optimizer	"You care about health but skip breakfast. Our 2-minute shake fixes the gap."
9	Survivorship Bias	Optimizer	"The 23% who succeeded all followed this pattern. The 77% did not."
10	Endowment Effect	Optimizer	Your workspace is already set up. Cancelling means losing everything you built.
11	Fundamental Attribution Error	Social	"It's not that you're bad at networking. Conferences are designed to exclude introverts."
12	Sunk Cost Fallacy	Optimizer	You've put 6 months into learning this skill. The next module is where it clicks.
13	Status Quo Bias	Filter	Works inside Slack. Your team won't even notice the switch.
14	False Consensus Effect	Social	"Most designers hate this tool. They just pretend they don't."
15	In-Group Favoritism	Social	The newsletter for founders who build in public - not sell courses.
16	Halo Effect	Optimizer	Designed by the same studio behind Apple's award-winning UI.
17	Hindsight Bias	Optimizer	"In 2022, we said no-code would eat SaaS. 3 years later - here's the data."
18	Backfire Effect	Filter	"You're right - cold email IS broken. That's exactly why we rebuilt the approach."
19	Bias Blind Spot	Filter	"If you're skeptical about these claims -- great. Let's look at the data."
20	Group Polarization	Social	Join 5,000 founders who are quitting the 9-to-5 this year.

Tone-of-Voice Switcher

Every output is tagged with one of 7 styles (or a hybrid with formula: 70% primary cadence + 30% secondary lexical markers). Each tone has a mandatory narrative minimum — e.g., warm-human requires a Named Person Story Arc in the first 3 sentences; luxe-minimal requires a sensory-physical anchor.

Style	Voice	Best for
bold-sell	Direct, urgent, high-energy	Flash sales, DTC, launches
expert-calm	Measured, analytical, data-rich	B2B, SaaS, consulting
rebel-edgy	Contrarian, disruptive, provocative	Youth brands, challengers, creators
warm-human	Empathetic, story-driven, vulnerable	Health, coaching, community
luxe-minimal	Sparse, polished, high-status	Premium, luxury, design
community-build	Inclusive, tribal, «we»-language	Community launches, membership
data-vivid	Numbers-driven, visual, concrete	Case studies, ROI pages, B2B decks

Bias Combinations (14 pre-built stacks)

Some biases multiply when combined. The skill provides 14 ready-to-use combos with conflict detection:

Combo	Biases	Use for
Trust Spiral	Authority → Social Proof → Confirmation → Endowment	Landing pages, sales pages
Urgency Engine	Loss Aversion → Social Proof → Scarcity	Flash sales, launches
Loyalty Loop	Confirmation → In-Group → Sunk Cost → Status Quo	Retention, upsells
Conversion Chain	Availability → Framing → Anchoring → Social Proof → Risk Reversal	Ads, free-to-paid
Cold-to-Warm Bridge	Availability → Framing → Authority → Social Proof	Cold audience → consideration
Trust-Repair Sequence	BBS(rev) → FAE(rev) → CogDiss → SQ(rev) → Reciprocity	Crisis, apology
Desire Escalator	Fear → Availability → Survivorship → Loss Aversion	Problem → solution
Objection Destroyer	Backfire → Anchoring → CogDiss → Risk Reversal	FAQ, skeptical audiences
Community Builder	In-Group → Group Polarization → False Consensus → Social Proof	Community launch
Premium Positioning	Halo → Anchoring → Authority → In-Group	Luxury, high-ticket
Launch Day Stack	Framing → Anchoring → Social Proof → Scarcity → Risk Reversal	Launch day
Lead Magnet Funnel	Reciprocity → Endowment → Authority → Sunk Cost	Freebie → nurture
Re-engagement Hook	Availability → Sunk Cost → In-Group → Loss Aversion	Lapsed customers
Micro-Content Burst	Framing + False Consensus + Social Proof (parallel)	Twitter/X, push notifications

Bias Selection Router

The skill doesn't guess which biases to use. SKILL.md contains an inlined Bias Selection Router - a single master table (50 rows) that maps audience × product × platform directly to a bias stack. Three steps: lookup → scenario override → category check. No multi-step decision-matrix traversal.

• Audience temperature (cold / warm / hot / lapsed / skeptical / stranger / defensive) ×
• Product type (low-B2C, mid-B2C, high-B2C, SaaS B2B, info-product, health, community) ×
• Platform (Twitter/X, LinkedIn, Instagram, TikTok, email, landing page, webinar, search ads)
• = Instant bias stack (3–5 biases, pre-validated for complementarity)
• Scenario Override - if user request matches a trigger (e.g., "launch", "apology"), biases are swapped per the Scenario Quick-Reference table.

Also includes a Bias Conflict Detector - 6 known bias conflicts with resolution strategies (e.g., Loss Aversion + Confirmation compete → sequence them: Fear first, Confirmation after solution).

Anti-Patterns (with Mechanical Detection Rules)

SKILL.md contains 12 anti-patterns inlined - each with a detection rule the LLM runs mechanically (grep-style) on its output before delivering:

#	Anti-Pattern	Detection Rule	Example Failure
1	Vague Social Proof	Search for "thousands","many" near customer claims → FAIL	"Thousands of satisfied customers"
2	Fear Without an Exit	Fear language without solution in same paragraph → FAIL	Threat with no immediate solution
3	Framing Without Anchoring	"Not X, but Y" without quantified old-way → FAIL	"This is a revolution" - no reference point
4	Authority Without Proof	"Studies show" without named source+year → FAIL	"Studies show..." - no source, no year
5	Transactional Reciprocity	Free value + pitch in same communication → FAIL	"Free ebook! Now buy."
6	Fake Scarcity	"Only X left" without explaining WHY → FAIL	Countdown timers that reset on reload
7	Premature Sunk Cost	"You've come this far" in first contact → FAIL	Sunk cost language in first paragraph
8	Empty In-Group	"Like-minded" without named identity → FAIL	"People like you" - no shared identity
9	Wrong-Source Halo	"Featured in [pub]" audience doesn't respect → FAIL	"As seen on Forbes" - audience doesn't respect Forbes
10	Insulting Confirmation	"Most people are wrong" → reader in "wrong" group → FAIL	Makes reader defensive, not engaged
11	Abstract Availability	"Imagine [scenario]" with <3 concrete sensory details → FAIL	No smells, sounds, times, names
12	Blaming Dissonance	"You say X but do Y" → blames reader, not system → FAIL	Shame → withdrawal, not action
13	Statistical-Only Fallacy	All social proof / authority claims are aggregate stats with zero named-person narrative → FAIL	Informs System 2, never activates System 1. Data without story = processed but not felt.

Post-Generation Verification: After writing, the LLM runs a mandatory 6-point mechanical checklist before delivering any output — checking for specific numbers, named sources, fear exits, scarcity explanations, system-blame, and human narrative depth (Named Person Story Arc).

What's inside

mindfluence/
├── SKILL.md                    ← Main instruction. Fully self-contained v2.1.
│                                  Inlined: Bias Selection Router, Scenario Quick-Reference,
│                                  Cultural Quick-Reference, 13 Anti-Patterns with detection rules
│                                  (incl. Statistical-Only Fallacy), Narrative Depth Requirements
│                                  (NPSA + Conversational Direct Address + Unexpected Detail),
│                                  14 Power Combinations, Bias Conflict Detector,
│                                  Post-Generation Verification (6 mechanical checks including HUMAN),
│                                  7 tones + hybrid formula with narrative minimums, 3-tier output format,
│                                  5 few-shot examples with NPSA annotations, Quality Rubric,
│                                  EXTERNAL REFERENCE FILES table with full GitHub URLs.
├── decision-matrix.md          ← Detailed audience×product×platform mapping. For deeper bias selection.
├── anti-patterns.md            ← 12 AI copywriting failures + detailed fixes + 30-sec audit checklist.
├── cultural-matrix.md          ← Full bias-by-bias cultural adaptation across 4 Hofstede dimensions.
├── scenarios/                  ← Task-specific playbooks (13 scenarios) — authoritative for complex tasks
│   ├── product-launch.md       ← Pre-launch → launch day → post-launch bias sequences
│   ├── social-media-post.md    ← X, LinkedIn, Instagram, Telegram, TikTok patterns
│   ├── landing-page.md         ← Full page bias map - hero to footer, section by section
│   ├── email-sequence.md       ← Welcome, nurture, cart abandonment, re-engagement
│   ├── webinar-warmup.md       ← Registration → live event (minute-by-minute) → post-webinar
│   ├── ad-campaign.md          ← Video, search, social, retargeting + A/B testing
│   ├── sales-page.md           ← 12-section psychological arc for high-ticket
│   ├── case-study.md           ← 6-part success story structure
│   ├── pricing-page.md         ← Anchoring architecture, decoy engineering
│   ├── cold-outreach.md        ← 5-line rule - no trust, no social proof
│   ├── crisis-response.md      ← 6-part apology structure - defensive bias engineering
│   ├── seo-brief.md            ← SEO brief integration, keyword density, humanization survivability
│   └── push-notifications.md   ← Push & SMS - lock screen psychology, 6 push types
├── examples/                   ← Annotated outputs (7 examples)
│   ├── social-post.md          ← LinkedIn post for freelancers (7 biases dissected)
│   ├── landing-hero.md         ← Hero section for premium meal delivery
│   ├── ad-script.md            ← 30s video ad for language app
│   ├── email-welcome.md        ← SaaS welcome email
│   ├── longform-article.md     ← Article intro with full bias analysis
│   ├── audit-example.md        ← Audit Mode - ad analyzed for biases + anti-patterns + fixes
│   └── optimize-hero.md        ← Optimize Mode - hero iterated from metrics
└── README.md

Example

Request: "Write a LinkedIn post about why freelancers undercharge."

Output: (annotated with [TONE], [BIASES ENGAGED], [TARGET ACTION], and [RATIONALE])

[TONE: warm-human × expert-calm]

I lost $23,000 before I understood this.

For 3 years, I charged clients based on what "felt right"...

→ [full example in examples/social-post.md]

Ethical Boundaries

Always:

• Use biases to get attention for genuinely valuable products
• Be transparent about what your product does and doesn't do
• Use factual, verifiable social proof

Never:

• Create false scarcity
• Fabricate testimonials or reviews
• Exploit fear to sell solutions that don't solve the fear
• Reinforce harmful beliefs
• Target vulnerable populations

The line: Would the customer, with full information and time to reflect, still choose your product? Yes = persuasion. No = manipulation.

Integration

MindFluence is designed for bi-directional integration with RankWise (SEO Content Engine) and HumanAI (Text Humanization Engine). The three skills form a complete content production pipeline: SEO structure → persuasive copy → human voice.

With RankWise (SEO Content Engine)

Pipeline: RankWise provides the SEO skeleton → MindFluence injects cognitive bias persuasion into it.

What MindFluence does:

• Uses the brief's heading structure (H1, H2, H3) as its skeleton - does not restructure
• Places the primary keyword in: H1, first paragraph, and at least one H2 (as specified in the brief)
• Keeps keyword density within the tone-specific cap (see table below)
• Preserves the brief's internal link plan (target URLs and suggested anchor text)
• Cross-references power words - avoids words that overlap with AI-burned-word lists

Keyword density caps by tone:

MindFluence Tone	Max Density	Notes
bold-sell	1.5%	Highest risk of over-stuffing. Cap strictly.
expert-calm	1.2%	Natural fit for informational content.
warm-human	1.2%	Conversational tone distributes keywords well.
rebel-edgy	1.5%	Contrarian framing may push keyword out - ensure first-150-word placement.
luxe-minimal	1.0%	Sparse copy - place keyword once prominently rather than forcing multiple instances.

Joint prompts:

• "RankWise SEO brief for [topic]. Keyword: [kw]. Then MindFluence from that brief. Expert-calm tone."
• "RankWise SEO brief for SaaS landing page. Then MindFluence landing page from that brief."

With HumanAI (Text Humanization)

Pipeline: MindFluence generates persuasive copy → HumanAI humanizes it without breaking the bias structure.

Tone mapping:

MindFluence Tone	HumanAI Tone
warm-human	human
expert-calm	expert
bold-sell	landing
rebel-edgy	social
luxe-minimal	expert

To preserve bias structure during humanization: use PIPELINE: cleanup → specificity → tone(skipped) → rhythm → proofread - this tells HumanAI to skip Stage 3 (tone) and keep the MindFluence voice.

Joint prompts:

• "Generate MindFluence copy (bold-sell tone), then HumanAI humanize it. EN."
• "MindFluence landing page for [product]. Then pass through HumanAI."

Bias-SEO Compatibility

Some cognitive biases have specific SEO interactions. See scenarios/seo-brief.md for the full table.

Bias	SEO Risk	Mitigation
Anchoring	Low	Numbers in opening satisfy C5 (number in title)
Loss Aversion	Low	"Stop losing X" - ensure keyword appears in first 150 words
Social Proof	Medium	Testimonials lack keywords → add keyword in framing sentences
Sunk Cost	Low	Time/effort framing doesn't conflict with SEO
Scarcity	Medium	"Limited", "Only X left" may push keyword out of first 150 words
Confirmation	Low	"You already know X" - fits keyword-adjacent language

Triple Pipeline (RankWise → MindFluence → HumanAI)

The complete production pipeline:

1. RankWise outputs a complete SEO brief with keyword plan, heading structure, meta drafts, internal link plan, and word count target.
2. MindFluence generates persuasive copy within the SEO skeleton, applying cognitive biases section by section.
3. HumanAI humanizes the copy, respecting: no deletion of keyword-containing headings, no deletion of internal link anchors, no cleanup of bias markers (social proof numbers, anchoring numbers), minimum word count preservation.
4. RankWise Audit performs final verification of the 49 SEO factors.

Joint triple prompt:

Triple pipeline:
1) RankWise SEO brief for [topic]. Keyword: [kw]. Language: [xx].
2) MindFluence generate from that brief. Tone: [expert-calm/warm-human/bold-sell].
3) HumanAI humanize the MindFluence output. Preserve SEO structure + bias markers.
4) RankWise audit the final result.

File Locations

Skill	Expected Location (OpenCode)	Expected Location (Claude Code)
RankWise	.opencode/skills/rankwise/	~/.claude/skills/rankwise/
MindFluence	.opencode/skills/mindfluence/	~/.claude/skills/mindfluence/
HumanAI	.opencode/skills/human-ai/	~/.claude/skills/human-ai/

---

Getting Started

Quick start (any LLM)

1. Copy the contents of SKILL.md
2. Paste as a system prompt / custom instructions
3. Type: "Write a post about [topic] for [audience] on [platform]"
4. For deeper context: the LLM can download additional files using the GitHub URLs in SKILL.md's EXTERNAL REFERENCE FILES section

The file is fully self-contained. No additional files needed for core operation. Bias Selection Router, Anti-Patterns with detection rules, Cultural Quick-Reference, few-shot examples - everything is inlined. If the LLM you're using can fetch external files, SKILL.md contains a full EXTERNAL REFERENCE FILES table with GitHub download URLs for all supplementary files (decision-matrix.md, anti-patterns.md, cultural-matrix.md, scenarios/, examples/).

With OpenCode (native)

Place the mindfluence/ folder into .opencode/skills/ (project) or ~/.config/opencode/skills/ (global). OpenCode auto-discovers SKILL.md via its frontmatter - no extra config needed.

With Claude Code

Copy SKILL.md into your project as CLAUDE.md, or place the full mindfluence/ folder under ~/.claude/skills/.

As a Custom GPT / Claude Project

Upload SKILL.md as the system prompt. For deeper context, add decision-matrix.md, anti-patterns.md, cultural-matrix.md, and files from scenarios/ and examples/ to the knowledge section. The file is self-contained for core operation, but the external files provide richer decision-matrix logic, detailed anti-pattern examples, and full scenario playbooks.

With any other LLM agent (Cursor, Windsurf, Copilot, Cline, Aider, etc.)

Copy the contents of SKILL.md into that agent's instruction file (.cursorrules, .windsurfrules, .github/copilot-instructions.md, .clinerules, CONVENTIONS.md, etc.). The prompt content is platform-agnostic.

Requirements

• Any capable LLM with a system prompt / custom instructions field (GPT, Claude, Gemini, DeepSeek, open-source models)
• One file: SKILL.md. Fully self-contained. All tables, detection rules, cultural data, and examples are inlined.
• Context window: 32K+ tokens for full v2.1 functionality (256K+ recommended for optimal quality)
• For deeper context: external files (decision-matrix.md, anti-patterns.md, cultural-matrix.md, scenarios/) preserved in the repo — full GitHub URLs in SKILL.md's EXTERNAL REFERENCE FILES section so LLMs working with one file can guide users to download them
• No API keys, no tools, no dependencies - pure prompt engineering
• Works in any language; adapts cultural references to the target market

Why "mindfluence"?

mind (the brain - cognitive biases live here) + influence (what persuasion actually does - shapes decisions without force).

---

License

MIT

---

GitHub Description

AI prompt-skill v2.1: 20 cognitive biases → high-converting marketing copy with guaranteed human narrative quality. Self-contained single-file with Bias Selection Router (audience×product×platform→instant bias stack), 13 anti-patterns with mechanical detection rules (incl. Statistical-Only Fallacy), Narrative Depth Requirements (Named Person Story Arc + Conversational Direct Address + Unexpected Detail), 14 power bias combinations, 7 tone styles + hybrid formula with narrative minimums, 3-tier output format (compact/standard/extended), 6-point post-generation verification (incl. HUMAN check), 12-region cultural quick-reference, and 5 inlined few-shot examples with NPSA annotations. Covers ads, landing pages, email sequences, social posts, webinars, launches, sales pages, case studies, pricing pages, cold outreach, push notifications, crisis response, and SEO brief integration.

GitHub Tags

prompt-engineering cognitive-biases marketing copywriting behavioral-economics psychology ai-prompts llm-skills persuasion neuromarketing conversion-optimization content-strategy system-prompt gpt claude self-contained bias-detection marketing-automation narrative-engineering human-quality