What problem does it solve? Gamification systems that grant XP, points, coins, reputation, achievements, streaks, likes, or referrals can be manipulated to emit rewards without the behavior they were designed to incentivize, inflating economies and rankings. This Skill gives an agent a systematic model to find where reward loops can be farmed, replayed, or self-awarded. ## Core Features & Use Cases - Reward Loop Modeling: Maps every reward as a TRIGGER → CONDITION → REWARD → REVERSAL loop and tests each phase for manipulation. - Abuse Vector Coverage: Hunts self-reward, multi-account referral farming, replay, concurrency farming, and bot automation, including the canonical ACTION → REWARD → REVERSE → ACTION → REWARD farming test. - Confidence-Graded Findings: Reports findings with evidence tiers from CONFIRMED (reproduced double reward) down to SPECULATIVE, plus false-positive checks like real reversals and unique constraints. - Use Case: While reviewing a social app, you discover users can react to their own posts and that unreacting does not remove the granted XP, enabling infinite XP farming through a react/unreact toggle. ## Quick Start Ask the agent to audit the reward and streak logic in this codebase for farming, self-reward, and replay abuse using the gamification audit.
