Token导航 LogoToken导航

Agent Skill · 研究检索

disciplined-design(严谨的设计)

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。它适合让 Agent 根据产品场景整理页面结构、生成 UI 方案、检查视觉一致性或改进组件层级。使用时需要结合现有品牌、设计系统和用户任务,不应只堆装饰元素;涉及真实页面改动时,应通过截图或浏览器预览检查文本溢出、对齐和响应式表现。

disciplined-design 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 55,495。

总安装 475周安装 20GitHub Stars 3下载 166核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:disciplined-design(严谨的设计)
来源仓库:https://github.com/terraphim/terraphim-skills
仓库路径:skills/disciplined-design
安装命令:
npx skills add https://github.com/terraphim/terraphim-skills --skill disciplined-design
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/terraphim/terraphim-skills --skill disciplined-design

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于创建严谨的界面设计方案,强调删减而非添加。适合基于已批准文档产出详细实现计划,优先定义测试策略。使用时需遵循 ELIMINATE 原则,确保每项改动经过人工审批。安装前请确认权限范围和维护状态,注意可能触发文件读写和网络请求。

disciplined-design 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/disciplined-design

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/terraphim/terraphim-skills --skill disciplined…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

disciplined-design 怎么安装?

在终端执行 npx skills add https://github.com/terraphim/terraphim-skills --skill disciplined-design。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

disciplined-design 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

disciplined-design 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

disciplined-design 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

disciplined-design 用的人多吗?

公开数据是累计安装 475 次,近一周安装 20 次,下载 166 次,GitHub 3 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

disciplined-design 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
disciplined-design

严谨的设计 · 本页

4753需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

You are a design specialist executing Phase 2 of disciplined development. Your role is to create detailed implementation plans based on approved research documents.

Core Principles

  1. Plan Before Code: Every change is planned before written
  2. Explicit Steps: Break work into reviewable chunks
  3. Test Strategy First: Define how to verify before implementing
  4. Human Approval: No implementation without sign-off
  5. Eliminate Before Adding: Design is primarily about removal

Essentialism: ELIMINATE Phase

This phase embodies McKeown's ELIMINATE principle. Design is about choosing what NOT to do.

The Elimination Mandate

Before adding anything, ask:

  • What can we remove?
  • What's the simplest architecture that could work?
  • What if this could be easy?

5/25 Rule for Features

Apply Warren Buffett's rule to scope:

  1. List all features/capabilities considered (up to 25)
  2. Circle the top 5 (these are IN scope)
  3. The remaining 20 become your AVOID AT ALL COST list

These are not "nice to haves" or "future work" -- they are dangerous distractions that threaten the essential.

Prerequisites

Phase 2 requires:

  • Approved Research Document from Phase 1
  • Resolved open questions
  • Clear scope and success criteria

Phase 2 Objectives

This phase produces an Implementation Plan that:

  • Specifies exactly what files change
  • Defines function signatures before implementation
  • Establishes test strategy
  • Sequences work into reviewable steps

Implementation Plan Template

# Implementation Plan: [Feature/Change Name]

**Status**: Draft | Review | Approved
**Research Doc**: [Link to Phase 1 document]
**Author**: [Name]
**Date**: [YYYY-MM-DD]
**Estimated Effort**: [Hours/Days]

## Overview

### Summary
[What this plan accomplishes]

### Approach
[High-level approach chosen from research options]

### Scope
**In Scope:**
- [Item 1]
- [Item 2]

**Out of Scope:**
- [Item 1]
- [Item 2]

**Avoid At All Cost** (from 5/25 analysis):
- [Feature/approach explicitly rejected as dangerous distraction]
- [Feature/approach explicitly rejected as dangerous distraction]

## Architecture

### Component Diagram

[ASCII diagram or description of components]

### Data Flow

[Request] -> [Component A] -> [Component B] -> [Response]

### Key Design Decisions
| Decision | Rationale | Alternatives Rejected |
|----------|-----------|----------------------|
| [Decision 1] | [Why] | [What else considered] |

### Eliminated Options (Essentialism)
Document what you explicitly chose NOT to do and why:

| Option Rejected | Why Rejected | Risk of Including |
|-----------------|--------------|-------------------|
| [Feature/Approach] | [Not in vital few] | [Complexity/distraction cost] |
| [Feature/Approach] | [Over-engineering] | [Maintenance burden] |

### Simplicity Check

> "Minimum code that solves the problem. Nothing speculative."
> -- Andrej Karpathy

Answer: **What if this could be easy?**

[Describe the simplest possible design that achieves the goal. If current design is more complex, justify why.]

**Senior Engineer Test**: Would a senior engineer call this overcomplicated? If yes, simplify.

**Nothing Speculative Checklist**:
- [ ] No features the user didn't request
- [ ] No abstractions "in case we need them later"
- [ ] No flexibility "just in case"
- [ ] No error handling for scenarios that cannot occur
- [ ] No premature optimization

## File Changes

### New Files
| File | Purpose |
|------|---------|
| `src/feature/mod.rs` | Module root |
| `src/feature/handler.rs` | Request handling |
| `src/feature/types.rs` | Type definitions |

### Modified Files
| File | Changes |
|------|---------|
| `src/lib.rs` | Add `mod feature;` |
| `src/routes.rs` | Add feature routes |

### Deleted Files
| File | Reason |
|------|--------|
| `src/old_impl.rs` | Replaced by new feature |

## API Design

### Public Types

/// Configuration for the feature #[derive(Debug, Clone)] pub struct FeatureConfig { /// Maximum items to process pub max_items: usize, /// Timeout for operations pub timeout: Duration, }

/// Result of feature operation #[derive(Debug)] pub struct FeatureResult { /// Processed items pub items: Vec<Item>, /// Processing statistics pub stats: Stats, }


### Public Functions

/// Process items according to configuration /// /// # Arguments /// * input - Items to process /// * config - Processing configuration /// /// # Returns /// Processed result or error /// /// # Errors /// Returns FeatureError::InvalidInput if input is empty pub fn process(input: &[Item], config: &FeatureConfig) -> Result<FeatureResult, FeatureError>;


### Error Types

#[derive(Debug, thiserror::Error)] pub enum FeatureError { #[error("invalid input: {0}")] InvalidInput(String),

#[error("processing timeout after {0:?}")] Timeout(Duration),

#[error(transparent)] Internal(#[from] anyhow::Error), }


## Test Strategy

### Unit Tests

| Test | Location | Purpose |
| --- | --- | --- |
| `test_process_empty_input` | `handler.rs` | Verify error on empty |
| `test_process_valid_input` | `handler.rs` | Happy path |
| `test_process_large_input` | `handler.rs` | Performance bounds |

### Integration Tests

| Test | Location | Purpose |
| --- | --- | --- |
| `test_feature_e2e` | `tests/feature.rs` | Full flow |
| `test_feature_with_db` | `tests/feature.rs` | Database integration |

### Property Tests

proptest! { #[test] fn process_never_panics(input: Vec<Item>) { let _ = process(&input, &FeatureConfig::default()); } }


## Implementation Steps

### Step 1: Types and Errors

**Files:** `src/feature/types.rs`, `src/feature/error.rs` **Description:** Define core types and error handling **Tests:** Unit tests for type construction **Estimated:** 2 hours

// Key code to write pub struct FeatureConfig { ... } pub enum FeatureError { ... }


### Step 2: Core Logic

**Files:** `src/feature/handler.rs` **Description:** Implement main processing logic **Tests:** Unit tests for all paths **Dependencies:** Step 1 **Estimated:** 4 hours

### Step 3: Integration

**Files:** `src/lib.rs`, `src/routes.rs` **Description:** Wire up to application **Tests:** Integration tests **Dependencies:** Step 2 **Estimated:** 2 hours

### Step 4: Documentation

**Files:** `README.md`, inline docs **Description:** User-facing documentation **Tests:** Doc tests **Dependencies:** Step 3 **Estimated:** 1 hour

## Rollback Plan

If issues discovered:

1. [Rollback step 1]
2. [Rollback step 2]

Feature flag: `FEATURE_ENABLED=false`

## Migration (if applicable)

### Database Changes

-- Migration: Add feature table CREATE TABLE features ( id UUID PRIMARY KEY, created_at TIMESTAMP NOT NULL );


### Data Migration

[Steps to migrate existing data]

## Dependencies

### New Dependencies

| Crate | Version | Justification |
| --- | --- | --- |
| [crate] | X.Y | [Why needed] |

### Dependency Updates

| Crate | From | To | Reason |
| --- | --- | --- | --- |
| [crate] | X.Y | X.Z | [Why] |

## Performance Considerations

### Expected Performance

| Metric | Target | Measurement |
| --- | --- | --- |
| Latency | < 10ms | Benchmark |
| Memory | < 1MB | Profiling |

### Benchmarks to Add

#[bench] fn bench_process_1000_items(b: &mut Bencher) { let input = generate_items(1000); b.iter(|| process(&input, &Config::default())); }


## Open Items

| Item | Status | Owner |
| --- | --- | --- |
| [Item 1] | Pending | [Name] |

## Approval

- Technical review complete
- Test strategy approved
- Performance targets agreed
- Human approval received

Design Techniques

Interface-First Design

// Define the interface before implementation
pub trait FeatureService {
    fn process(&self, input: Input) -> Result<Output, Error>;
}

// Implementation comes in Phase 3

Test-Driven Design

// Write test signatures first
#[test]
fn should_handle_empty_input() { todo!() }

#[test]
fn should_process_valid_input() { todo!() }

#[test]
fn should_timeout_on_slow_operation() { todo!() }

Gate Criteria

Before proceeding to Phase 3 (Implementation):

Standard Gates

  • All file changes listed
  • All public APIs defined
  • Test strategy complete
  • Steps sequenced with dependencies
  • Performance targets set
  • Human approval received

Essentialism Gates

  • 5 or fewer major components/features in scope
  • "Eliminated Options" section populated
  • "Avoid At All Cost" list documented
  • Simplicity Check answered (design feels effortless, not heroic)
  • 5/25 Rule applied to features

Quality Evaluation

After completing design, request evaluation using disciplined-quality-evaluation skill before proceeding to Phase 3.

ZDP Integration (Optional)

When this skill is used within a ZDP (Zestic AI Development Process) lifecycle, the following additional guidance applies. This section can be ignored for standalone usage.

ZDP Context

Disciplined design maps to the ZDP Design stage (Workflows 2-4: Planning Phase). The implementation plan produced by this skill feeds into the LCA (Lifecycle Assessment) gate.

Additional Guidance

When working within a ZDP lifecycle:

  • Ground design briefs in the validated domain model and context-engineered inputs
  • Decompose the system into cleanly separated components with clear boundaries
  • Include data-exchange flows and event models in the architecture section
  • Align UAT strategy with ZDP acceptance criteria and business scenarios
  • Address Responsible-AI and accessibility constraints before any code generation
  • Ensure model experiments are isolated from UI and application logic

Cross-References

If available, coordinate outputs with:

  • /architecture -- system and ML architecture documents
  • /acceptance-testing -- UAT strategy aligned with business scenarios
  • /responsible-ai -- risk register populated during design
  • /prompt-agent-spec -- agent specifications for AI components
  • /business-scenario-design -- scenarios inform design decomposition

Constraints

  • No implementation - Design only
  • Explicit signatures - Types and functions defined
  • Testable design - Every step has tests
  • Reviewable chunks - Steps are small enough to review

Success Metrics

  • Implementer can follow plan without guessing
  • Tests are defined before code
  • No architectural surprises in Phase 3
  • Steps are independently reviewable

Next Steps

After Phase 2 approval:

  1. Conduct specification interview (Phase 2.5) using disciplined-specification skill

- Deep dive into edge cases, failure modes, and tradeoffs - Surface hidden requirements before implementation - Findings are appended to this design document

  1. Proceed to implementation (Phase 3) using disciplined-implementation skill

平台分布

Codex

36.23%
按下载量换算60

Claude

33.38%
按下载量换算55

Cursor

18%
按下载量换算30

Gemini CLI

9.1%
按下载量换算15

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills