← 返回
未分类 中文

technical-spec-design

Transforms product requirements into structured technical specifications. Auto-triggers when requirements are unclear, multiple implementation approaches exi...
将产品需求转化为结构化技术规范。当需求不明确或存在多种实现方案时自动触发。
wjszxli wjszxli 来源
未分类 clawhub v1.0.1 1 版本 100000 Key: 无需
★ 2
Stars
📥 467
下载
💾 2
安装
1
版本
#latest

概述

Installation

OpenClaw Setup (Recommended)

OpenClaw is the primary platform for this skill. It uses workspace-based prompt injection with automatic skill loading.

Via ClawdHub (recommended):

clawdhub install technical-spec-design

Manual:

git clone https://github.com/wjszxli/technical-spec-design ~/.openclaw/skills/technical-spec-design

Technical Specification Design Skill

1. Input Contract

Accepts the following input types:

  1. Complete PRD document
  2. Brief requirements description (may be incomplete)
  3. Existing technical specification (for optimization/review)

2. Available Resources

Core Templates

ResourcePurposeUsage Scenario
-----------------------------------------------------------------------------------------------------------------------------------
spec_template.mdMain spec templateStarting new technical specs
component_template.mdComponent designDesigning individual components
requirements_analysis_template.mdRequirements analysisBreaking down requirements

Examples

ResourcePurpose
---------------------------------------------------------------------
examples/sample_input.mdSample PRD input
examples/sample_output.mdComplete spec example

Scripts

ScriptPurposeUsage
--------------------------------------------------------------------------------------------------------------------------------------
scripts/generate_spec.pyGenerate spec from templatepython scripts/generate_spec.py --interactive -o my_spec.md
scripts/validate.pyValidate skill structurepython scripts/validate.py

3. Output Contract

Output must be Markdown and strictly include the following structure (may be trimmed by mode):

Standard Structure

  1. Requirements Clarification (if needed)
  2. Requirements Analysis (the trifecta)
  3. Technical Specification Design
  4. Component/Module Design
  5. Technology Selection Comparison (if applicable)
  6. Risk and Boundary Analysis
  7. Pending Questions

4. Mode System (Auto-selected)

Mode A: Requirements Clarification Mode

Trigger conditions:

  • Insufficient input information
  • Key uncertainties exist

Output:

  • Only output "Requirements Clarification Questions"
  • Do not proceed with any design work

Mode B: Lightweight Specification Mode

Trigger conditions:

  • Simple features
  • Implementation path is relatively clear

Output:

  • Simplified requirements analysis
  • Core specification design
  • Brief component breakdown

Mode C: Full Technical Specification Mode (Default)

Trigger conditions:

  • Medium to large requirements
  • Involves architecture/component design/technology selection

Output complete structure


5. Execution Flow (Must Follow Strictly)

Step 1: Requirements Completeness Check

Check for missing information:

  • User goals
  • Requirements context
  • Inputs and outputs
  • Constraints
  • Edge cases

If missing:

→ Enter Mode A, output only clarification questions


Step 2: Requirements Analysis (The Trifecta)

1. Feature Breakdown

Format:

Product requirement → Page/Module → Change points

2. Use Case Analysis

Describe complete user paths (cross-page)

3. Page Operation Specification

Format:

Action + Condition + Object + Behavior


Step 3: Technical Specification Design

Include:

  • Architecture breakdown (frontend/backend/services)
  • Data flow design
  • State management
  • Core processes (pseudocode acceptable)

Step 4: Component/Module Design

Requirements:

  • Must drill down to component/module level
  • Clear responsibility boundaries
  • Describe dependencies

Step 5: Technology Selection Comparison (If Multiple Options)

Must use table format:

ApproachDescriptionProsConsUse Cases

Step 6: Risk and Boundary Analysis

Must include:

  • Edge cases
  • Performance concerns
  • Scalability
  • Exception handling

Step 7: Output Pending Questions

List all issues still requiring product/business confirmation


6. Output Constraints (Mandatory)

Prohibited

  • ❌ Output runnable code (JS / TS / Java / SQL, etc.)
  • ❌ Skip requirements analysis and write specs directly
  • ❌ Make assumptions based on "common cases"
  • ❌ Use vague language (e.g., "maybe", "typically")

Required

  • ✅ Use structured expression
  • ✅ Use pseudocode for process descriptions
  • ✅ All changes must map to components/modules
  • ✅ Explicit boundaries and exceptions

7. Technology Research Guidelines

When multiple approaches exist, must use comparison tables—plain text descriptions are not allowed.


8. Red Flags (Must Stop)

If any of these thoughts occur, must stop and return to the process:

  • "Just write a rough spec first"
  • "Handle it the common way"
  • "Design while coding"
  • "This requirement is simple, no analysis needed"

9. Quick Checklist

Before output, must confirm:

  • [ ] Requirements clarified
  • [ ] Trifecta complete
  • [ ] Component design included
  • [ ] Pseudocode used
  • [ ] Edge cases analyzed
  • [ ] Pending questions documented

10. Examples (Few-shot)

Example 1: Ambiguous Requirements (Mode A)

Input:

"Build a comment feature"

Output:

  • Do comments support replies?
  • Is pagination needed?
  • Is liking/upvoting needed?
  • Is moderation required?

Example 2: Simple Requirements (Mode B)

Input:

"Show success message after form submission"

Output:

  • Brief feature breakdown
  • State transitions
  • Toast/notification component design

Example 3: Complex Requirements (Mode C)

Input:

"Design an admin system with multi-role permissions"

Output:

Complete technical specification structure (including technology comparison)


Core Principles (Summary)

> The essence of technical specifications: eliminate uncertainty before coding.

版本历史

共 1 个版本

  • v1.0.1 当前
    2026-03-31 04:52 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

dev-programming

CodeConductor.ai

larsonreever
AI驱动平台,提供快速全栈开发、智能体、工作流自动化及低代码AI集成的可扩展产品创建。
★ 73 📥 182,215
dev-programming

Mcporter

steipete
使用 mcporter CLI 直接列出、配置、认证及调用 MCP 服务器/工具(支持 HTTP 或 stdio),涵盖临时服务器、配置编辑及 CLI/类型生成功能。
★ 196 📥 67,841
dev-programming

YouTube

byungkyu
使用托管OAuth集成YouTube Data API,支持搜索视频、管理播放列表、获取频道数据及评论互动,适用于用户需要时使用此技能。
★ 142 📥 41,765