technical-research

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Technical Research Skill

概述

Overview

此技能提供专业的技术调研能力，将模糊的技术问题转化为结构化、可执行的调研流程，输出可直接指导实现的调研文档。

This skill provides professional technical research capabilities, transforming vague technical problems into structured, executable research processes, and outputting research documents that can directly guide implementation.

核心原则

Core Principles

1. 简洁至上 (Concise is Key)

1. Concise is Key

上下文窗口是公共资源，只包含 Claude 不知道的信息
每个信息点都要问："Claude 真的需要这个解释吗？"
偏好简洁示例，避免冗长说明

Context window is a shared resource, only include information that Claude doesn't know
For each information point, ask: "Does Claude really need this explanation?"
Prefer concise examples, avoid lengthy explanations

2. 渐进式披露 (Progressive Disclosure)

2. Progressive Disclosure

采用三级加载系统：

Metadata - 始终在上下文 (~100 词)
SKILL.md body - 技能触发后加载 (<500 行)
Bundled resources - 按需加载 (无限制)

Adopt a three-level loading system:

Metadata - Always in context (~100 words)
SKILL.md body - Loaded after skill is triggered (<500 lines)
Bundled resources - Loaded on demand (unlimited)

3. 设定适当的自由度 (Set Appropriate Degrees of Freedom)

3. Set Appropriate Degrees of Freedom

自由度	使用场景	示例
高 (文本指令)	多种方法有效、依赖上下文决策	"选择最适合的数据库"
中 (伪代码/参数化脚本)	有偏好模式、可接受变化	"使用此模板生成 API"
低 (具体脚本、少参数)	操作脆弱、一致性关键	"执行此迁移脚本"

Degree of Freedom	Use Case	Example
High (Text Instructions)	Multiple methods are effective, context-dependent decision making	"Choose the most suitable database"
Medium (Pseudocode/Parameterized Scripts)	Preference patterns exist, changes are acceptable	"Use this template to generate API"
Low (Specific Scripts, Few Parameters)	Operations are fragile, consistency is critical	"Execute this migration script"

调研流程

Research Process

阶段 1: 理解需求 (5-10 分钟)

Phase 1: Understand Requirements (5-10 minutes)

目标：将模糊问题转化为具体调研目标

关键问题：

markdown

1. 业务背景是什么？(解决什么问题)
2. 技术约束有哪些？(预算、团队技能、时间)
3. 成功标准是什么？(性能指标、成本上限)
4. 已有技术栈是什么？(需要兼容的系统)

输出：调研目标声明

markdown

undefined

Objective: Transform vague problems into specific research objectives

Key Questions:

markdown

1. What is the business background? (What problem are we solving?)
2. What are the technical constraints? (Budget, team skills, time)
3. What are the success criteria? (Performance metrics, cost ceiling)
4. What is the existing tech stack? (Systems that need to be compatible with)

Output: Research Objective Statement

markdown

undefined

调研目标

Research Objectives

评估 X 技术在 Y 场景下的适用性
对比 A/B/C 三种方案的优缺点
确定实现 Z 功能的最优路径

undefined

Evaluate the applicability of technology X in scenario Y
Compare the advantages and disadvantages of solutions A/B/C
Determine the optimal path to implement function Z

undefined

阶段 2: 信息收集 (15-30 分钟)

Phase 2: Information Gathering (15-30 minutes)

并行收集策略：

┌─────────────────────────────────────────────────────┐
│                  信息收集矩阵                        │
├──────────────┬──────────────┬───────────────────────┤
│ 官方文档     │ GitHub 示例   │ 社区讨论              │
│ - API 参考    │ - 实际用法    │ - 痛点/陷阱           │
│ - 最佳实践   │ - 生产代码    │ - 替代方案            │
└──────────────┴──────────────┴───────────────────────┘

收集清单：

官方文档核心章节
GitHub 高星项目实现
StackOverflow 高频问题
技术博客/视频分享
竞品/替代方案对比

Parallel Collection Strategy:

┌─────────────────────────────────────────────────────┐
│                  Information Gathering Matrix        │
├──────────────┬──────────────┬───────────────────────┤
│ Official Docs│ GitHub Examples│ Community Discussions │
│ - API Reference │ - Practical Usage │ - Pain Points/Pitfalls │
│ - Best Practices │ - Production Code │ - Alternative Solutions │
└──────────────┴──────────────┴───────────────────────┘

Collection Checklist:

Core chapters of official documentation
High-star GitHub project implementations
High-frequency StackOverflow questions
Technical blogs/video shares
Competitor/alternative solution comparisons

阶段 3: 分析评估 (20-40 分钟)

Phase 3: Analysis and Evaluation (20-40 minutes)

评估维度：

python

undefined

Evaluation Dimensions:

python

undefined

技术评估框架

Technology Evaluation Framework

evaluation = { "功能性": { "功能覆盖率": "0-100%", "扩展性": "高/中/低", "定制能力": "强/中/弱" }, "非功能性": { "性能": "基准测试结果", "可靠性": "SLA/故障率", "安全性": "认证/审计", "可维护性": "文档质量/社区活跃度" }, "成本": { "许可成本": "免费/商业/订阅", "学习成本": "陡峭/平缓", "运维成本": "高/中/低" }, "风险": { "技术成熟度": "实验/成熟/衰退", "供应商锁定": "高/中/低", "社区可持续性": "强/中/弱" } }

undefined

evaluation = { "Functionality": { "Function Coverage": "0-100%", "Scalability": "High/Medium/Low", "Customization Capability": "Strong/Medium/Weak" }, "Non-Functionality": { "Performance": "Benchmark test results", "Reliability": "SLA/Failure rate", "Security": "Certification/Audit", "Maintainability": "Documentation quality/Community activity" }, "Cost": { "License Cost": "Free/Commercial/Subscription", "Learning Cost": "Steep/Gentle", "Operation and Maintenance Cost": "High/Medium/Low" }, "Risk": { "Technology Maturity": "Experimental/Mature/Declining", "Vendor Lock-in": "High/Medium/Low", "Community Sustainability": "Strong/Medium/Weak" } }

undefined

阶段 4: 输出文档 (15-25 分钟)

Phase 4: Document Output (15-25 minutes)

标准输出结构：

markdown

undefined

Standard Output Structure:

markdown

undefined

[技术名称] 调研报告

[Technology Name] Research Report

概述 (1-2 句)

Overview (1-2 sentences)

[技术的核心价值和适用场景]

[Core value and applicable scenarios of the technology]

核心发现 (Bullet Points)

Key Findings (Bullet Points)

[关键发现 1]
[关键发现 2]
[关键发现 3]

[Key Finding 1]
[Key Finding 2]
[Key Finding 3]

技术细节

Technical Details

[实现原理、架构设计、关键代码]

[Implementation principles, architecture design, key code]

优缺点分析

Pros and Cons Analysis

优点	缺点
...	...

Pros	Cons
...	...

适用场景

Applicable Scenarios

✅ 适合：[场景列表]
❌ 不适合：[场景列表]

✅ Suitable for: [Scenario List]
❌ Not suitable for: [Scenario List]

最佳实践

Best Practices

[生产环境建议]

[Production environment recommendations]

参考资料

References

[链接列表]

undefined

[Link List]

undefined

文档规范

Document Specifications

YAML Frontmatter (必需)

YAML Frontmatter (Required)

所有调研文档必须包含：

yaml

---
title: 文档标题
date: YYYY-MM-DD
topic: [ai-llm|frontend|backend|database|devops|tools]
tags: [标签 1, 标签 2]
status: [draft|complete|archived]
related: [相关文档路径]
---

All research documents must include:

yaml

---
title: Document Title
date: YYYY-MM-DD
topic: [ai-llm|frontend|backend|database|devops|tools]
tags: [Tag 1, Tag 2]
status: [draft|complete|archived]
related: [Related Document Path]
---

文件命名

File Naming

格式：
```
YYYY-MM-DD-topic.md
```
位置：
```
research/{topic}/
```

示例：

research/ai-llm/2026-02-22-rag-architecture.md

Format:
```
YYYY-MM-DD-topic.md
```
Location:
```
research/{topic}/
```

Example:

research/ai-llm/2026-02-22-rag-architecture.md

标签体系

Tag System

主题标签：

#ai

#llm

#frontend

#react

#backend

#api

#database

#sql

#nosql

#devops

#docker

#k8s

状态标签：

#draft

#complete

#needs-review

#archived

优先级标签：

#high

#medium

#low

Topic Tags:

#ai

#llm

#frontend

#react

#backend

#api

#database

#sql

#nosql

#devops

#docker

#k8s

Status Tags:

#draft

#complete

#needs-review

#archived

Priority Tags:

#high

#medium

#low

调研模式

Research Modes

模式 1: 技术选型调研

Mode 1: Technology Selection Research

触发场景：需要在多个技术方案中选择

流程：

1. 明确选择标准 (权重分配)
2. 列出候选方案 (3-5 个)
3. 逐项对比评分
4. 计算加权总分
5. 给出推荐方案

输出模板：见

references/templates/selection-matrix.md

Trigger Scenario: Need to choose among multiple technical solutions

Process:

1. Clarify selection criteria (weight allocation)
2. List candidate solutions (3-5)
3. Compare and score item by item
4. Calculate weighted total score
5. Provide recommended solution

Output Template: See

references/templates/selection-matrix.md

模式 2: 实现方案调研

Mode 2: Implementation Scheme Research

触发场景：需要实现特定功能

流程：

1. 功能分解 (子任务列表)
2. 收集实现模式 (3+ 种)
3. 评估每种模式
4. 推荐最优实现
5. 提供代码示例

输出模板：见

references/templates/implementation-guide.md

Trigger Scenario: Need to implement specific functions

Process:

1. Function decomposition (sub-task list)
2. Collect implementation patterns (3+ types)
3. Evaluate each pattern
4. Recommend optimal implementation
5. Provide code examples

Output Template: See

references/templates/implementation-guide.md

模式 3: 架构设计调研

Mode 3: Architecture Design Research

触发场景：需要设计系统架构

流程：

1. 需求分析 (功能/非功能)
2. 约束识别 (技术/业务)
3. 模式选择 (架构风格)
4. 组件设计 (模块划分)
5. 接口定义 (API 设计)

输出模板：见

references/templates/architecture-design.md

Trigger Scenario: Need to design system architecture

Process:

1. Requirement analysis (functional/non-functional)
2. Constraint identification (technical/business)
3. Pattern selection (architecture style)
4. Component design (module division)
5. Interface definition (API design)

Output Template: See

references/templates/architecture-design.md

模式 4: 问题排查调研

Mode 4: Troubleshooting Research

触发场景：需要解决技术问题

流程：

1. 问题复现 (最小化)
2. 根因分析 (5 Why)
3. 方案收集 (社区/文档)
4. 方案评估 (风险/成本)
5. 解决实施 + 验证

输出模板：见

references/templates/troubleshooting.md

Trigger Scenario: Need to solve technical problems

Process:

1. Problem reproduction (minimization)
2. Root cause analysis (5 Why)
3. Solution collection (community/documentation)
4. Solution evaluation (risk/cost)
5. Solution implementation + verification

Output Template: See

references/templates/troubleshooting.md

质量检查清单

Quality Checklist

常见反模式

Common Anti-Patterns

❌ 过度收集

❌ Over-Collection

表现：收集大量信息但无法提炼核心发现

解决：设定时间盒，强制输出 3 个核心发现

Performance: Collect a lot of information but cannot extract key findings

Solution: Set time box, force output of 3 key findings

❌ 缺乏结论

❌ Lack of Conclusion

表现：罗列信息但不给出推荐

解决：必须包含"推荐方案"和"理由"

Performance: List information but do not provide recommendations

Solution: Must include "Recommended Solution" and "Reasons"

❌ 忽略约束

❌ Ignoring Constraints

表现：推荐方案不考虑实际约束

解决：明确列出技术/业务约束

Performance: Recommended solution does not consider actual constraints

Solution: Clearly list technical/business constraints

❌ 代码不可运行

❌ Unrunnable Code

表现：示例代码未经验证

解决：至少验证核心代码片段

Performance: Sample code is unverified

Solution: At least verify core code snippets

工具使用

Tool Usage

信息收集工具

Information Gathering Tools

bash

undefined

bash

undefined

官方文档

Official Documentation

webfetch <url>

GitHub 代码搜索

GitHub Code Search

grep_app_searchGitHub <pattern>

技术社区

Technical Community

task(subagent_type="librarian", prompt="...")

undefined

task(subagent_type="librarian", prompt="...")

undefined

文档生成工具

Document Generation Tools

bash

undefined

bash

undefined

使用模板

Use Template

cp docs/templates/research-template.md research/ai-llm/YYYY-MM-DD-topic.md

更新索引

Update Index

编辑 SUMMARY.md 添加新文档

undefined

Edit SUMMARY.md to add new document

undefined

技能资源

Skill Resources

脚本 (

scripts/

)

Scripts (

scripts/

)

```
validate_research.py
```
- 验证调研文档完整性
```
generate_summary.py
```
- 自动生成 SUMMARY.md 索引

```
validate_research.py
```
- Validate research document integrity
```
generate_summary.py
```
- Automatically generate SUMMARY.md index

参考 (

references/

)

References (

references/

)

```
templates/
```
- 各类调研文档模板
```
checklists/
```
- 质量检查清单
```
patterns/
```
- 常见调研模式

```
templates/
```
- Various research document templates
```
checklists/
```
- Quality checklists
```
patterns/
```
- Common research patterns

资产 (

assets/

)

Assets (

assets/

)

```
evaluation-matrix.xlsx
```
- 技术评估矩阵模板
```
decision-tree.png
```
- 技术选型决策树

```
evaluation-matrix.xlsx
```
- Technology evaluation matrix template
```
decision-tree.png
```
- Technology selection decision tree

迭代改进

Iterative Improvement

使用技能后，记录：

调研效率：从问题到输出的时间
决策质量：推荐方案的实际效果
文档复用：后续参考频率

基于反馈更新：

SKILL.md 中的流程指导
references/ 中的模板
scripts/ 中的自动化脚本

版本：1.0.0
作者：Research Team
最后更新：2026-02-22

After using the skill, record:

Research Efficiency: Time from problem to output
Decision Quality: Actual effect of recommended solution
Document Reusability: Subsequent reference frequency

Update based on feedback:

Process guidance in SKILL.md
Templates in references/
Automation scripts in scripts/

Version: 1.0.0
Author: Research Team
Last Updated: 2026-02-22