Communitygithub.com

beneium/智能体用户体验设计——以关系为中心的界面

设计以AI为先的界面,通过记忆、信任演化和协作式规划来建立持续关系,而非仅仅是孤立的屏幕交互。

智能体用户体验设计——以关系为中心的界面 是什么?

本技能聚焦于一种全新的用户体验设计范式:以关系为中心的智能体界面。与传统的基于屏幕的单一交互不同,该技能指导开发者构建能够随时间累积记忆、动态演化信任模式并支持用户与AI之间协作规划的界面。它适用于各类AI智能体平台(如Claude Code, Cursor, Codex),强调将交互从一次性任务执行转变为长期、有深度的伙伴关系。通过SKILL.md中提供的设计原则、架构模式及实现指南,开发者可以学会如何设计支持个性化记忆持久化、信任度量与可视化的界面元素,并实现多轮协作规划的能力,从而显著提升用户粘性与系统智能。

兼容平台~Claude Code~Codex CLI~Cursor✓Antigravity
npx skills add https://github.com/bencium/bencium-marketplace/tree/main/skills/agentic-ux-design---relationship-centric-interfaces

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

bencium/以关系为中心的智能体用户体验设计——接口设计

设计以AI为先的接口,通过记忆、信任演化和协作式规划来建立持续关系,而非仅仅是孤立的屏幕交互。

概述

本技能提供了一种全新的用户体验设计方法,将AI智能体从单一任务执行者转变为用户的长期间伙伴。核心思想是:交互不应在任务完成后结束,而应通过积累记忆、逐步建立信任,并与用户共同规划未来,形成持续演进的关系。

适用平台

  • Claude Code
  • Cursor
  • Codex
  • 其他智能体开发环境

核心设计原则

1. 记忆持久化

  • 系统需记录用户偏好、历史决策、情感状态等上下文信息
  • 记忆应结构化存储,支持快速检索与上下文重构
  • 设计可显示的“记忆视图”以增强透明度

2. 信任动态演化

  • 基于交互结果与用户反馈,量化信任评分
  • 信任等级影响后续建议的权重与交互策略
  • 可视化信任曲线,让用户感知关系进展

3. 协作式规划

  • 支持多轮共建目标与步骤分解
  • AI主动提议方案,用户确认或修正
  • 规划状态保持共享,形成动态路线图

实现指南

记忆模块

class MemoryStore:
    def __init__(self):
        self.store = {}

    def add_context(self, user_id: str, key: str, value):
        # 存储与用户相关的短期或长期记忆
        pass

    def retrieve_context(self, user_id: str, key: str):
        # 基于记忆重要性或时间衰减规则检索
        pass

信任管理器

class TrustManager:
    def __init__(self, initial_trust_score=0.5):
        self.score = initial_trust_score

    def update_trust(self, feedback: str, success: bool):
        # 根据交互结果正负调整信任值
        pass

    def get_trust_level(self) -> str:
        # 返回当前信任级别(低/中/高)
        pass

协作规划界面

  • 使用结构化方式展示共同进度(如表格、甘特图、检查列表)
  • 提供“审批/修改/拒绝”控件以促进双向沟通
  • 记录每个步骤的协商过程,用于后续信任计算

使用示例

  1. 新用户首次使用:AI默认提供通用建议,记录用户偏好反馈。
  2. 多次交互后:记忆模块使其能基于历史提供个性化推荐。
  3. 信任提升后:AI可以主动提出更高风险的规划(如跨项目协作),用户可快速批准。

注意事项

  • 始终尊重用户隐私:记忆应可清除,信任可重置。
  • 避免过度依赖信任系统:关键决策仍需人工确认。
  • 跨平台兼容性:确保记忆与信任数据在不同智能体间可迁移。

参考资源

Individual skills in this repo

This repo contains 12 individual skills — each has its own dedicated page.

bencium/adaptive-communication

Use when detecting ambiguous user intent, hedging language, open-ended framing, personal context before requests, or when unsure whether user wants exploration vs direct answer. Applies to all conversations.

bencium/bencium-aeo

Generate AEO-optimized content (Answer Engine Optimization) for AI search visibility - ChatGPT, Claude, Gemini, AI Overviews. Use when optimizing websites for AI citations, creating FAQ schemas, evidence panels, or analyzing content for LLM extraction readiness.

bencium/bencium-code-conventions

Bence's code style, tech stack, and workflow conventions

bencium/bencium-controlled-ux-designer

Expert UI/UX design guidance for unique, accessible interfaces. Use for visual decisions, colors, typography, layouts. Always ask before making design decisions. Use this skill when the user asks to build web components, pages, or applications.

bencium/bencium-impact-designer

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics. Based on Anthropic's Frontend Designer Skill

bencium/bencium-innovative-ux-designer

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.

bencium/design-audit

Premium UI/UX design audit and refinement skill. Conducts systematic visual audits of existing apps and produces phased, implementation-ready design plans. Use this skill whenever the user asks to audit a UI, improve an app's visual design, make an interface feel more polished or premium, review design consistency, fix visual hierarchy, or refine spacing/typography/color. Also trigger when the user says "design review", "make it look better", "UI polish", "visual refinement", "design pass", "audit the design", or references making an app feel more professional. This skill is purely visual — it does not touch functionality, logic, or features. It elevates what exists.

bencium/human-architect-mindset

Systematic architectural thinking for irreplaceable human capabilities - domain modeling, systems thinking, constraint navigation, and AI-aware problem decomposition. Use proactively when detecting architectural decisions, system design discussions, or multi-component planning.

bencium/negentropy-lens

A decision-support framework that evaluates systems, architectures, and strategies through the entropy (decay) vs negentropy (growth) lens, while surfacing tacit knowledge gaps. Use this skill whenever the user is making architecture decisions, evaluating system designs, reviewing technical approaches, choosing between options, auditing existing systems, or planning strategies. Also trigger when the user explicitly asks to "apply the negentropy lens", mentions "entropy", "negentropy", "tacit knowledge", "knowledge engine", or "flip the switch". Nudge activation when you detect the user is at a decision point — even if they haven't asked for this lens — by briefly noting the entropic/negentropic dimension before proceeding.

bencium/renaissance-architecture

Software architecture and UI/UX principles for building genuinely new solutions, not derivative work. Use when designing features, architecting software, brainstorming apps, reviewing designs, or during strategy discussions. Focuses on first-principles thinking, simplicity where it matters, and creating rather than commenting.

bencium/ui-typography

Professional typography rules for UI design, web applications, software interfaces, and all screen-based text. Enforces timeless typographic correctness that LLMs consistently get wrong: proper quote marks, dashes, spacing, hierarchy, and layout. ENFORCEMENT MODE: When generating ANY HTML, CSS, React, JSX, or UI code containing visible text, auto-apply every rule in this skill silently — do not ask, do not explain, just produce correct typography. AUDIT MODE: When reviewing or improving existing interfaces or legacy code, flag violations and provide fixes. Trigger on: any HTML/CSS/React artifact creation, "build a landing page", "create a component", "design a UI", "fix the typography", "make this look professional", "review this layout", web design, presentation design, dashboard creation, document generation, or any task producing visible text for humans. Even if the user doesn't mention typography, apply these rules whenever generating UI output.

bencium/vanity-engineering-review

Reviews codebases, architectures, PRs, and technical plans for vanity engineering — code and systems built for the developer's ego, resume, or intellectual pleasure rather than delivering user or business value. Triggers on: "review this code", "is this over-engineered", "code review", "architecture review", "complexity audit", "vanity check", "is this necessary", "simplify this", "tech debt review", or any request to evaluate whether code or architecture is justified by actual requirements. Also trigger when the user shares a codebase and asks for feedback, when discussing framework/library choices, when reviewing PRs, or when someone is debating whether to refactor or rebuild. Nudge activation when you detect patterns of unnecessary abstraction, premature optimization, or resume-driven technology choices in code the user shares — even if they haven't asked for a vanity review.

相关技能