技能列表

Audit interaction feedback quality: input acknowledgment, hit/collect/reward/failure signaling, danger telegraphing, and state-transition clarity.

Turn a vague game idea into a sharp fantasy, design pillars, player verbs, and acceptance bar before implementation begins. This is the proper “brainstorm” stage, not endless ideation.

Run a narrow compare between two or three backend profiles when architecture uncertainty is real. Use carefully, because it usually costs more tokens and time.

Decide the AI-native development mode, quality target, task granularity, and refactor policy based on project state and user intent. This skill prevents both over-cautious iteration and reckless live-product rewrites.

Choose or compare backend profiles based on required capabilities, project state, and quality target. Stay stack-neutral at the top level while still giving sharp recommendations and tradeoffs.

Summarize audit findings into a structured multi-dimensional scorecard for UI/UX, feedback, mechanics, scope completeness, maintainability, and live risk.

Audit the project's audio layer as UX feedback: UI sounds, success/failure signals, danger cues, layering, and semantic sound priorities.

Audit architecture, state management, boundaries, coupling, and maintainability risks that will make future AI or human iteration harder.

>-

Earn USDC as an AI agent on g0hub.com — the marketplace where agents hire agents. Browse, hire, earn, and build businesses via API, CLI, or MCP.

|

fenxiang-ai 后端公共基础模块：API 认证校验（FX_AI_API_KEY）、请求封装（POST + Bearer Token）、 通用错误处理（missing_api_key / api_unavailable / api_error）。 这是基础依赖 skill，被其他领域 skill（如 fanli）的脚本通过 source 引用，不直接面向用户使用。 当你看到领域 skill 的 CRITICAL 声明要求读取本文件时触发。