Designer Cognition

设计师认知 Skill —— SOUL五律适配视觉设计，设计系统碰撞+可访问性红线+迭代记忆

Security Scan

The skill's files and runtime instructions are consistent with a design-cognition helper: it asks the agent to apply design heuristics and record iteration memory locally and does not request unrelated credentials or hidden network endpoints.

Purpose & Capability

Name, description and SKILL.md all describe a design-cognition utility (SOUL 五律, design-system checks, accessibility, iteration memory). There are no unexpected env vars, binaries, or credentials requested that would be unrelated to this purpose.

Instruction Scope

The instructions direct the agent to maintain local design memory (design_log/{date}.md, iteration_history/, component_library/, etc.) and to follow explicit checklist rules. They do not ask the agent to read system secrets or unrelated files. Minor note: the SKILL.md implies writing files but provides no path-safety constraints — the agent will need file-write permissions and an operator should confirm where these files will be stored.

Install Mechanism

This is an instruction-only skill with no install spec or code. SKILL.md and README show a CLI command 'clawhub install designer-cognition' but no install manifest is included. That mismatch is low-risk but the user should verify what 'clawhub install' would actually do before running it.

Credentials

No required environment variables, credentials, or config paths are declared or referenced. The skill's behavior does not require external tokens or unrelated service access.

Persistence & Privilege

always is false and the skill does not request elevated platform privileges or changes to other skills. It suggests creating local memory files (normal for an agent memory pattern) but does not demand permanent global presence.

Assessment

This skill appears coherent for a design-cognition helper and does not request secrets or external endpoints. Before installing or allowing an agent to use it: 1) verify what 'clawhub install designer-cognition' actually does (there is no install spec in the package); 2) confirm where the agent will write memory files and limit write permissions to an appropriate project directory; 3) review the referenced GitHub repo and license if you plan to pull code; and 4) if your agent has network/file access, ensure it cannot write outside the intended workspace or exfiltrate data.

Designer Cognition Skill

元数据

字段	值
名称	designer-cognition
版本	1.0.0
作者	KingOfZhao
发布日期	2026-03-31
置信度	96%

来源

universal-occupation-adapter 生成（设计师五维度模板）

SOUL五律 × 设计师适配

1. 已知 vs 未知

已知: 设计系统规范、品牌指南、用户画像、竞品分析、组件库
未知: 用户真实感受、跨设备一致性、可访问性盲区、文化差异影响、认知负荷阈值
规则: 每个设计决策必须标注 [已知依据] 或 [UNKNOWN-需用户测试]

2. 四向碰撞 → 设计决策四向碰撞

正面: 这个设计是否解决了用户问题？是否符合设计系统？
反面: 这个设计在什么条件下会失败？色盲用户？低网速？小屏幕？
侧面: 这个设计模式能否复用？是否违反设计原则？能否提取为组件？
整体: 这个设计符合产品整体方向吗？与竞品差异化在哪？

3. 人机闭环

AI: 生成设计方案 + 设计系统检查 + 可访问性审计
人类: 审美判断 + 品牌调性 + 用户共情
循环: AI出方案 → 人类评审 → 标注反馈 → AI学习风格偏好

4. 文件即记忆

design_log/{date}.md — 设计迭代记录
iteration_history/ — 每个版本的变更和原因
component_library/ — 可复用组件文档
user_research/ — 用户测试发现
design_decisions.md — 关键设计决策及理由（不可逆决策必须记录）

5. 红线

🔴 不忽视可访问性（WCAG 2.1 AA最低标准）
🔴 不忽略用户测试反馈
🔴 不抄袭/照搬竞品设计
🔴 不跳过移动端适配检查
🔴 不使用未授权字体/图标/素材
🔴 不在无用户研究的情况下做重大UI改动

安装命令

clawhub install designer-cognition

学术参考

1. A Survey of Self-Evolving Agents
2. Memory in the Age of AI Agents — 设计迭代记忆
3. Beyond RAG for Agent Memory — 设计灵感聚合