Technical Research Skill

Overview

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.

Core Principles

1. Concise is Key

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. Progressive Disclosure

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. Set Appropriate Degrees of Freedom

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"

Research Process

Phase 1: Understand Requirements (5-10 minutes)

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

## Research Objectives
- 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

Phase 2: Information Gathering (15-30 minutes)

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

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

Evaluation Dimensions:

python

# Technology Evaluation Framework
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"
    }
}

Phase 4: Document Output (15-25 minutes)

Standard Output Structure:

markdown

# [Technology Name] Research Report

## Overview (1-2 sentences)
[Core value and applicable scenarios of the technology]

## Key Findings (Bullet Points)
- [Key Finding 1]
- [Key Finding 2]
- [Key Finding 3]

## Technical Details
[Implementation principles, architecture design, key code]

## Pros and Cons Analysis
| Pros | Cons |
|------|------|
| ... | ... |

## Applicable Scenarios
- ✅ Suitable for: [Scenario List]
- ❌ Not suitable for: [Scenario List]

## Best Practices
[Production environment recommendations]

## References
[Link List]

Document Specifications

YAML Frontmatter (Required)

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]
---

File Naming

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

Example:

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

Tag System

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

Research Modes

Mode 1: Technology Selection Research

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

Mode 2: Implementation Scheme Research

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

Mode 3: Architecture Design Research

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

Mode 4: Troubleshooting Research

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

Quality Checklist

Before outputting the research document, check:

Are key findings clear (3-5 bullet points)
Is the pros and cons analysis balanced (unbiased)
Are code examples runnable (verified)
Are references complete (traceable)
Are applicable scenarios clear (not misleading)
Is YAML frontmatter correct
Is the document placed in the correct directory

Common Anti-Patterns

❌ Over-Collection

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

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

❌ Lack of Conclusion

Performance: List information but do not provide recommendations

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

❌ Ignoring Constraints

Performance: Recommended solution does not consider actual constraints

Solution: Clearly list technical/business constraints

❌ Unrunnable Code

Performance: Sample code is unverified

Solution: At least verify core code snippets

Tool Usage

Information Gathering Tools

bash

# Official Documentation
webfetch <url>

# GitHub Code Search
grep_app_searchGitHub <pattern>

# Technical Community
task(subagent_type="librarian", prompt="...")

Document Generation Tools

bash

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

# Update Index
Edit SUMMARY.md to add new document

Skill Resources

Scripts (

scripts/

)

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

References (

references/

)

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

Assets (

assets/

)

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

Iterative Improvement

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

technical-research

NPX Install

Tags

SKILL.md Content (Chinese)

Technical Research Skill

Overview

Core Principles

1. Concise is Key

2. Progressive Disclosure

3. Set Appropriate Degrees of Freedom

Research Process

Phase 1: Understand Requirements (5-10 minutes)

Phase 2: Information Gathering (15-30 minutes)

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

Phase 4: Document Output (15-25 minutes)

Document Specifications

YAML Frontmatter (Required)

File Naming

Tag System

Research Modes

Mode 1: Technology Selection Research

Mode 2: Implementation Scheme Research

Mode 3: Architecture Design Research

Mode 4: Troubleshooting Research

Quality Checklist

Common Anti-Patterns

❌ Over-Collection

❌ Lack of Conclusion

❌ Ignoring Constraints

❌ Unrunnable Code

Tool Usage

Information Gathering Tools

Document Generation Tools

Skill Resources

Scripts (
`scripts/`
)

References (
`references/`
)

Assets (
`assets/`
)

Iterative Improvement

technical-research

NPX Install

Tags

SKILL.md Content (Chinese)

Technical Research Skill

Overview

Core Principles

1. Concise is Key

2. Progressive Disclosure

3. Set Appropriate Degrees of Freedom

Research Process

Phase 1: Understand Requirements (5-10 minutes)

Phase 2: Information Gathering (15-30 minutes)

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

Phase 4: Document Output (15-25 minutes)

Document Specifications

YAML Frontmatter (Required)

File Naming

Tag System

Research Modes

Mode 1: Technology Selection Research

Mode 2: Implementation Scheme Research

Mode 3: Architecture Design Research

Mode 4: Troubleshooting Research

Quality Checklist

Common Anti-Patterns

❌ Over-Collection

❌ Lack of Conclusion

❌ Ignoring Constraints

❌ Unrunnable Code

Tool Usage

Information Gathering Tools

Document Generation Tools

Skill Resources

Scripts (scripts/)

References (references/)

Assets (assets/)

Iterative Improvement

Scripts (
`scripts/`
)

References (
`references/`
)

Assets (
`assets/`
)