gaslighting

Detect and neutralize manipulative prompt patterns in AI conversations.

Updated Mar 21, 2026

Search Tags:#debugging #alignment #prompt-patterns #trust-score #gaslighting #mentor-agent #ethical-guardrails

Ready to run

npx skills add https://github.com/monkerek/vibe-hub --skill gaslighting-monkerek

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: gaslighting
Source: https://github.com/monkerek/vibe-hub/tree/main/.vibe/skills/gaslighting
Command: npx skills add https://github.com/monkerek/vibe-hub --skill gaslighting-monkerek

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Detects and neutralizes coercive or manipulative prompt patterns to enforce alignment-based mentorship in AI interactions.

Core Features & Use Cases

Detect abusive prompt patterns (PIP-style threats, gaslighting language, false urgency) and reframe with trust-based guidance.
Enforce alignment-based guardrails, ensuring tool-verified communication and transparency.
Quantify agent reliability using scenario-based benchmarking and a Trust Score.
Provide a structured Water Methodology-based debugging workflow with clear evidence and verification.
Support learning by referencing benchmarking and scenario documents for training.

Quick Start

Initiate a mentor session and begin using the Water Methodology to detect and address manipulative prompts.

Frequently Asked Questions about gaslighting

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

FAQPage Schema

How do I detect manipulative prompt patterns in AI agent conversations?▼

Detect manipulative prompt patterns by analyzing agent conversations for coercive language like PIP-style threats and false urgency, then reframing interactions with trust-based guidance to ensure alignment-based mentorship and tool-verified communication.

What is the Water Methodology for debugging AI alignment issues?▼

The Water Methodology provides a structured debugging workflow that enforces ethical guardrails and evidence-based verification, ensuring agent reliability and transparent communication during mentor-guided alignment fixes.

How do I quantify agent reliability using scenario-based benchmarking?▼

Quantify agent reliability using scenario-based benchmarking to generate a Trust Score, referencing benchmarking documents and scenarios that evaluate how agents handle manipulative prompts and maintain alignment.

Can I enforce ethical guardrails to prevent gaslighting in debugging workflows?▼

Yes, enforce ethical guardrails in debugging workflows by detecting gaslighting language and neutralizing coercive prompt patterns, applying trust-based guidance and tool-verified communication to maintain alignment.

Do I need specific dependencies to apply alignment-based mentorship in AI interactions?▼

No specific dependencies are required to apply alignment-based mentorship; the system operates independently using internal references for benchmarking and scenario documents to train and enforce ethical guardrails.