Build an Anti-Manipulation AI Prompt Toolkit Powered by the Principles of Influence
You'll translate the detection logic behind Cialdini's six principles of influence into a reusable AI prompt toolkit covering at least 5 real-world scenarios—live-stream shopping, course purchases, investment advice, job-offer outreach, social-media ads, and more—so you can call up an AI diagnostic any time before an important decision to expose the persuasion mechanisms at play, get ready-made counter-scripts, and receive clear-headed decision guidance.
Final work
An "Anti-Manipulation AI Prompt Toolkit" based on *Influence*
Estimated time
1–2 hr
Submitted
…
Your final work
Purpose:Package the detection logic of the six principles of influence into on-demand AI system prompts and user-input templates, so that when you face a persuasion scenario, AI can help you diagnose the trigger mechanism, supply counter-scripts, and generate clear-headed decision guidance.
Parts:
At least 5 scene-specific AI system prompts (each containing the six-principles inspection instruction)
A user-input template for each scene (with scene-description placeholders)
A description of the expected output format for each scene
A 'When to Use Which Tool' usage manual
A complete set of sample anti-manipulation counter-script outputs
Use cases:
· Real-time diagnosis of live-stream shopping and e-commerce promotions
· Assisted review for course or training purchase decisions
· Distinguishing genuine from manipulative investment and wealth-management pitches
· Vetting job offers and headhunter outreach
· Deconstructing workplace or personal-relationship PUA scripts
Pick a topic
Pick the topic closest to you, or write a custom one when you submit.
Personal Life
Learning & Growth
Family & Parenting
Work & Projects
Communication & Relationships
Tools you'll use from the book
Six-Principles Detection System Prompt
Encode the recognition signals for the six principles of influence (reciprocity, commitment & consistency, social proof, liking, authority, scarcity) into an AI role instruction so the AI plays the role of a 'persuasion-mechanism analyst.'
How to use it here:
Write an AI system prompt for your toolkit that includes: a role definition (persuasion-mechanism analyst), a six-principles recognition checklist, and an output format (principle name → detected script signal → risk level → recommended decision action).
Boundaries:
The system prompt should not ask AI to 'refuse on your behalf' or 'decide for you.' It should provide diagnostic results and options; final decision authority belongs to you.
Anti-Manipulation Counter-Script Generator
Based on the specific trigger mechanisms detected, ask AI to generate counter-scripts or pause-scripts you can say on the spot (e.g., 'I need 24 hours to think about this').
How to use it here:
Add a 'script generation' module to your system prompt: when you paste a persuasive pitch, AI first diagnoses the mechanism, then offers 2–3 ready-to-use response lines (no need to be confrontational—just buy yourself time to think).
Boundaries:
Scripts should remain polite and non-confrontational. The goal is to buy yourself decision time, not to fight back. Avoid letting AI generate aggressive responses.
Decision-Matrix Prompt
Transform the defensive framework from Chapter 8 of *Influence* into a structured decision-matrix prompt: input a persuasion scenario → output a 'signal vs. genuine need' comparison table.
How to use it here:
Design a prompt that asks AI to output a simple table: 'Which principle was triggered in this scenario' vs. 'If I remove that trigger, do I still want this?' — helping you separate the impulse from the real need.
Boundaries:
The matrix should focus on whether the decision still holds after removing the trigger, not on dismissing all persuasion as invalid.
Devil's Advocate Prompt
Have AI play 'devil's advocate' and surface 3 reasons *against* buying or agreeing, forcing you to actively seek positive evidence yourself.
How to use it here:
Add a 'forced-rebuttal mode' prompt to your toolkit: you input the decision you're leaning toward, and AI outputs only 3 reasons against it (no supporting reasons), making you think through why you actually want to do it.
Boundaries:
Devil's advocate mode should be a supplementary tool for 'I'm already convinced but want one more check'—it shouldn't be the default flow for every decision.
Third-Party Perspective Prompt
Have AI evaluate the current persuasion scenario from the viewpoint of 'a rational friend,' simulating the mindset shift of 'If a close friend faced this situation, what would you tell them?'
How to use it here:
Design a prompt where AI acts as 'your rational friend': rather than just analyzing principles, it speaks in first-person conversational style—'If I were you, here's how I'd see this…'—neutralizing the self-justification bias.
Boundaries:
The third-party perspective prompt is best suited for emotional decision scenarios (e.g., when the liking or reciprocity principle is activated). It's not necessary for purely rational decisions.
Work rules
Your work MUST include
Complete AI system prompts for at least 5 scenes (each prompt must include six-principles inspection instructions and output-format definition)
A paired user-input template for each system prompt (with concrete placeholders, e.g., "[Paste the persuasive script here]")
A complete 'sample conversation' for at least 1 scene—showing real input and AI output (the prompt alone is not enough; prove it actually works)
A usage manual explaining which tool to use in which scenario and the recommended order of use
At least 3 principle-specific recognition signals cited from the book (drawn from the original text, not generic summaries)
Your work CANNOT just be
Don't just restate the six principles without providing executable AI prompts
Don't write prompts that ask AI to decide whether to buy on your behalf—decision-making authority must stay with you
Don't let AI generate scripts that attack or humiliate the other party
Don't only provide a prompt framework without at least one complete sample output
Don't spread yourself too thin across all 9 scenes—do 5 thoroughly instead
AI can help you here
Round 1: Help me choose scenes and design the first system prompt
When to use: You're just getting started, don't know which scene to tackle first, or aren't sure how to write an effective AI system prompt.
I'm working through "{{book title}}" on the "{{route name}}" project, and my goal is to build an Anti-Manipulation AI Prompt Toolkit.
My situation:
[Describe the persuasion scenarios you encounter most often—e.g., I impulse-buy during live streams, I'm easily swayed by authority endorsements before buying courses, a financial advisor keeps pushing products on me.]
Please help me do two things:
First, based on my situation, recommend 5 of the following scenes that I should prioritize, and explain why:
- Live-stream sales-script diagnosis kit
- Course-purchase decision assistant
- Investment-advice anti-manipulation tool
- Job-offer authenticity checker
- Social-media ad diagnosis kit
- Parenting-pitch translator
- Anti-PUA tool
- Influencer content authenticity detector
- Refusing-sales-pressure kit
Second, for the first scene you recommend, write a complete AI system prompt that must include:
1. A role definition (persuasion-mechanism analyst)
2. A checklist of recognition signals for Cialdini's six principles (reciprocity, commitment & consistency, social proof, liking, authority, scarcity)
3. An output-format definition (format: "Principle identified → Specific script signal → Risk level (low/medium/high) → Recommended decision action")
4. One boundary statement (when this tool is not sufficient)
Also provide the paired user-input template (with placeholders).
Yellow placeholders need you to fill in before using the AI.
AI can help you organize ideas, but cannot make final judgments for you. Don't let AI fabricate experiences, cases, or misleading content.
Round 2: Help me expand the toolkit and generate sample outputs
When to use: You've finished system prompts for 1–2 scenes and now need to extend the remaining scenes, or you want to verify how effective a prompt actually is.
I'm working through "{{book title}}" on the "{{route name}}" project. I have completed system prompts for the following scenes:
My topic: {{topic}}
Completed prompts (optional):
[Paste your existing system prompts here]
Please help me do two things:
First, write a new system prompt for the scene below (same requirements as before: role definition + six-principles checklist + output format + usage boundary):
[Enter the name of the next scene you want to build, e.g., "Course-purchase decision assistant"]
Second, generate a complete sample conversation for one of my existing system prompts:
- Sample input (simulate a real persuasion scenario the user might enter)
- Sample output (in the format defined in the prompt, showing what AI should output)
The sample must cover at least 3 different principles of influence, and the recognition signals must be specific (don't just say "scarcity principle was used"—quote the exact script).
Yellow placeholders need you to fill in before using the AI.
AI can help you organize ideas, but cannot make final judgments for you. Don't let AI fabricate experiences, cases, or misleading content.
Round 3: Help me review the full toolkit and write the usage manual
When to use: You've completed all scene prompts and are ready to organize everything into your final submission.
I'm submitting my Shufang Island project work.
Book: "{{book title}}"
Project route: {{route name}}
My topic: {{topic}}
My draft work:
{{draft work}}
Please help me do three things:
First, check whether my toolkit meets the following criteria:
1. Does it include complete system prompts for at least 5 scenes?
2. Does each prompt have a role definition + six-principles checklist + output format?
3. Is there at least 1 complete sample conversation (real input + full AI output)?
4. Does the output format in the prompts keep decision-making authority with the user (rather than having AI decide for the user)?
5. Do at least 3 principle references cite specific recognition-signal details from the book?
Second, if I haven't written a usage manual, please draft one for me: explain which tool to use for which scenario, the recommended order of use, and caveats (including when *not* to use this toolkit).
Third, give an overall evaluation of the full toolkit:
- What's done well
- What must be added or revised
- What could be strengthened
- The one last thing to do before submitting
Yellow placeholders need you to fill in before using the AI.
AI can help you organize ideas, but cannot make final judgments for you. Don't let AI fabricate experiences, cases, or misleading content.