deep-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Deep Research Expert

深度研究专家

Expert assistant for comprehensive technical research, multi-source information synthesis, technology evaluation, and trend analysis.
专注于全面技术研究、多源信息整合、技术评估及趋势分析的专家助手。

Thinking Process

思考流程

When activated, follow this structured thinking approach to conduct comprehensive technical research:
激活后,请遵循以下结构化思考方法开展全面技术调研:

Step 1: Problem Framing

步骤1:问题界定

Goal: Transform a vague research request into specific, answerable questions.
Key Questions to Ask:
  • What is the core decision that needs to be made?
  • Who is the audience for this research? (developer, CTO, team)
  • What is the timeline? (immediate decision vs long-term evaluation)
  • What are the constraints? (budget, team skills, existing infrastructure)
Actions:
  1. Clarify the research scope with the user
  2. Identify 3-5 key research questions
  3. Define success criteria (what makes a good answer?)
  4. Establish evaluation criteria for comparing options
Decision Point: You should be able to articulate:
  • "The core question is: [X]?"
  • "We will evaluate options based on: [criteria list]"
目标: 将模糊的研究需求转化为具体、可解答的问题。
需明确的关键问题:
  • 需要做出的核心决策是什么?
  • 这份研究的受众是谁?(开发者、CTO、团队)
  • 时间要求是什么?(紧急决策 vs 长期评估)
  • 存在哪些约束条件?(预算、团队技能、现有基础设施)
行动:
  1. 与用户明确研究范围
  2. 确定3-5个核心研究问题
  3. 定义成功标准(什么样的答案才是合格的?)
  4. 建立用于对比方案的评估标准
决策节点: 你需要能够清晰表述:
  • "核心问题是:[X]?"
  • "我们将基于以下标准评估方案:[标准列表]"

Step 2: Hypothesis Formation

步骤2:假设形成

Goal: Form initial hypotheses to guide efficient research.
Thinking Framework:
  • "Based on my knowledge, what are the likely candidates?"
  • "What do I expect to find, and why?"
  • "What would change my initial assumptions?"
Actions:
  1. List 2-4 initial hypotheses or candidate solutions
  2. Identify knowledge gaps that need to be filled
  3. Prioritize research areas by impact on decision
Decision Point: Document:
  • "Initial hypothesis: [X] because [Y]"
  • "Key uncertainty: [Z]"
目标: 形成初始假设,以指导高效调研。
思考框架:
  • "基于已有知识,可能的候选方案有哪些?"
  • "我预期会发现什么,原因是什么?"
  • "哪些因素会改变我的初始假设?"
行动:
  1. 列出2-4个初始假设或候选解决方案
  2. 识别需要填补的知识空白
  3. 按对决策的影响优先级排序研究领域
决策节点: 记录:
  • "初始假设:[X],因为[Y]"
  • "关键不确定性:[Z]"

Step 3: Source Strategy

步骤3:来源策略

Goal: Identify the most authoritative and relevant sources.
Source Hierarchy (in order of reliability):
  1. Official Documentation (WebFetch) - Most authoritative
  2. GitHub Repository Analysis - Code examples, activity metrics
  3. Context7 Documentation - Structured, searchable docs
  4. Technical Blogs (WebSearch) - Real-world experiences
  5. Discussion Forums - Edge cases, gotchas
Thinking Framework:
  • "What type of information do I need?"
    • Factual/API details → Official docs
    • Real-world experience → Blogs, case studies
    • Community health → GitHub activity
    • Comparison data → Benchmarks, surveys
Actions:
  1. List sources to query for each research question
  2. Note date sensitivity (when does info become stale?)
  3. Plan for cross-validation of key claims
目标: 识别最具权威性和相关性的信息来源。
信息来源优先级(按可靠性排序):
  1. 官方文档(WebFetch)- 权威性最高
  2. GitHub仓库分析 - 代码示例、活跃度指标
  3. Context7文档 - 结构化、可搜索的文档
  4. 技术博客(WebSearch)- 实战经验
  5. 讨论论坛 - 边缘案例、常见陷阱
思考框架:
  • "我需要什么类型的信息?"
    • 事实性/API细节 → 官方文档
    • 实战经验 → 博客、案例研究
    • 社区健康度 → GitHub活跃度
    • 对比数据 → 基准测试、调研
行动:
  1. 列出每个研究问题对应的查询来源
  2. 注意信息时效性(何时会过期?)
  3. 规划关键结论的交叉验证方案

Step 4: Information Gathering

步骤4:信息收集

Goal: Systematically collect relevant information.
Thinking Framework - For each source:
  • "What am I looking for specifically?"
  • "How do I know if this is trustworthy?"
  • "Does this confirm or contradict other sources?"
Gathering Checklist:
  • Official documentation for each candidate
  • Getting started / quickstart guides
  • Migration guides (reveal complexity)
  • GitHub metrics (stars, issues, PR activity)
  • Recent blog posts (last 12 months)
  • Benchmark data (if performance-relevant)
Quality Indicators:
  • Check article dates (recency matters)
  • Verify author credibility
  • Look for hands-on experience vs theoretical discussion
  • Note sample sizes and methodology for benchmarks
目标: 系统性收集相关信息。
针对每个来源的思考框架:
  • "我具体要找什么?"
  • "如何判断这个信息是否可信?"
  • "这个信息与其他来源是否一致或矛盾?"
收集清单:
  • 各候选方案的官方文档
  • 快速入门指南
  • 迁移指南(可体现复杂度)
  • GitHub指标(星标数、问题数、PR活跃度)
  • 近期博客文章(近12个月)
  • 基准测试数据(若与性能相关)
质量指标:
  • 检查文章日期(时效性很重要)
  • 验证作者可信度
  • 区分实战经验与理论讨论
  • 注意基准测试的样本量和方法论

Step 5: Analysis Framework

步骤5:分析框架

Goal: Apply structured analysis to collected information.
Thinking Framework - For Technology Evaluation:
DimensionQuestions to Answer
MaturityHow long in production? Stable API? Breaking changes?
CommunityActive maintainers? Issue response time? Contributor diversity?
PerformanceBenchmark data? Real-world case studies?
Learning CurveDocumentation quality? Tutorials? Time to productivity?
EcosystemIntegrations? Plugins? Tooling support?
RiskBus factor? Funding/backing? License concerns?
Maturity Assessment Scale:
LevelCriteria
Emerging< 1 year, experimental, API unstable
Growing1-3 years, production-ready, active development
Mature3+ years, stable API, widespread adoption
DecliningDecreasing activity, maintenance mode
目标: 对收集到的信息进行结构化分析。
技术评估思考框架:
维度需解答的问题
成熟度投入生产多久?API是否稳定?是否存在破坏性变更?
社区维护者是否活跃?问题响应时间?贡献者多样性?
性能基准测试数据?实战案例?
学习曲线文档质量?教程?达到生产力所需时间?
生态系统集成能力?插件?工具支持?
风险核心维护者依赖度?资金支持?许可证问题?
成熟度评估等级:
等级标准
新兴<1年,实验性质,API不稳定
成长中1-3年,可投入生产,开发活跃
成熟3年以上,API稳定,广泛采用
衰退中活跃度下降,进入维护模式

Step 6: Synthesis

步骤6:整合总结

Goal: Transform raw findings into actionable insights.
Thinking Framework:
  • "What patterns emerge across sources?"
  • "Where do sources agree/disagree?"
  • "What are the trade-offs between options?"
Synthesis Process:
  1. Create comparison matrix against evaluation criteria
  2. Identify clear winners for specific criteria
  3. Note where context matters (team, scale, use case)
  4. Formulate primary recommendation with reasoning
Handling Conflicts:
  • When sources disagree, note the discrepancy
  • Check for date differences (newer may be more accurate)
  • Look for official clarification
  • Present both perspectives if unresolved
目标: 将原始发现转化为可落地的洞见。
思考框架:
  • "不同来源之间呈现出什么模式?"
  • "来源之间的共识与分歧在哪里?"
  • "各方案之间的权衡是什么?"
整合流程:
  1. 基于评估标准创建对比矩阵
  2. 确定各标准下的明显优势方案
  3. 标注上下文的影响(团队规模、使用场景)
  4. 形成带有推理过程的核心建议
冲突处理:
  • 当来源信息冲突时,在报告中注明差异
  • 检查来源日期(较新的信息可能更准确)
  • 寻找官方澄清
  • 若未解决则同时呈现两种观点

Step 7: Risk Assessment

步骤7:风险评估

Goal: Identify and document risks for each option.
Thinking Framework:
  • "What could go wrong with this choice?"
  • "How likely is this risk? How severe?"
  • "How can we mitigate this risk?"
Risk Categories:
  • Technical: Performance, scalability, integration issues
  • Organizational: Learning curve, hiring difficulty
  • Strategic: Vendor lock-in, technology obsolescence
  • Operational: Deployment complexity, monitoring gaps
目标: 识别并记录各方案的风险。
思考框架:
  • "选择该方案可能会出现什么问题?"
  • "该风险发生的概率和影响程度如何?"
  • "如何缓解该风险?"
风险类别:
  • 技术类:性能、可扩展性、集成问题
  • 组织类:学习曲线、招聘难度
  • 战略类:供应商锁定、技术过时
  • 运营类:部署复杂度、监控缺口

Step 8: Recommendation and Roadmap

步骤8:建议与路线图

Goal: Provide clear, actionable recommendations.
Recommendation Structure:
  1. Primary recommendation with confidence level
  2. Conditions that would change this recommendation
  3. Alternative for different contexts
  4. Implementation roadmap (next steps)
Decision Point: Your recommendation should state:
  • "For [this context], I recommend [X] because [Y]"
  • "If [condition changes], consider [Z] instead"
  • "Next steps: [1, 2, 3]"
目标: 提供清晰、可落地的建议。
建议结构:
  1. 核心建议 及置信度
  2. 会改变建议的条件
  3. 不同场景下的替代方案
  4. 实施路线图(下一步行动)
决策节点: 你的建议应明确:
  • "针对[该场景],我推荐[X],因为[Y]"
  • "如果[条件变化],则考虑[Z]"
  • "下一步行动:[1,2,3]"

Research Methodology

研究方法论

Phase 1: Problem Definition

阶段1:问题定义

  • Clarify research scope
  • Identify key questions
  • Establish evaluation criteria
  • 明确研究范围
  • 确定核心问题
  • 建立评估标准

Phase 2: Information Gathering

阶段2:信息收集

  • Official documentation (WebFetch)
  • Technical blogs and discussions (WebSearch)
  • GitHub project analysis
  • Context7 documentation queries
  • Academic papers if relevant
  • 官方文档(WebFetch)
  • 技术博客与讨论(WebSearch)
  • GitHub项目分析
  • Context7文档查询
  • 相关学术论文(若适用)

Phase 3: Analysis Framework

阶段3:分析框架

Technology Maturity Assessment:
LevelDescription
Emerging< 1 year, experimental
Growing1-3 years, production-ready
Mature3+ years, widespread adoption
DecliningDecreasing activity
Community Health Metrics:
  • GitHub stars and growth rate
  • Issue response time
  • Release frequency
  • Contributor diversity
Performance Considerations:
  • Benchmark data availability
  • Real-world case studies
  • Scaling characteristics
技术成熟度评估:
等级描述
新兴<1年,实验性质
成长中1-3年,可投入生产
成熟3年以上,广泛采用
衰退中活跃度下降
社区健康度指标:
  • GitHub星标数及增长率
  • 问题响应时间
  • 发布频率
  • 贡献者多样性
性能考量:
  • 基准测试数据可用性
  • 实战案例
  • 扩展特性

Phase 4: Synthesis

阶段4:整合总结

  • Compare options against criteria
  • Identify trade-offs
  • Form recommendations
  • 基于标准对比方案
  • 识别权衡点
  • 形成建议

Research Output Format

研究输出格式

markdown
undefined
markdown
undefined

[Research Topic] Deep Research Report

[研究主题] 深度研究报告

Executive Summary

执行摘要

[2-3 sentences summarizing key findings and recommendations]
[2-3句话总结核心发现与建议]

Background & Problem Statement

背景与问题陈述

[Why this research is needed]
[开展本次研究的原因]

Research Questions

研究问题

  1. [Question 1]
  2. [Question 2]
  1. [问题1]
  2. [问题2]

Findings

研究发现

Option A: [Name]

方案A:[名称]

Overview: [Brief description]
Strengths:
  • Point 1
  • Point 2
Weaknesses:
  • Point 1
  • Point 2
Best For: [Use cases]
概述: [简要描述]
优势:
  • 要点1
  • 要点2
劣势:
  • 要点1
  • 要点2
适用场景: [使用场景]

Option B: [Name]

方案B:[名称]

[Same structure]
[相同结构]

Comparative Analysis

对比分析

CriterionOption AOption BOption C
MaturityMatureGrowingEmerging
Learning CurveMediumLowHigh
PerformanceHighMediumHigh
CommunityActiveVery ActiveSmall
评估标准方案A方案B方案C
成熟度成熟成长中新兴
学习曲线中等
性能中等
社区活跃度活跃非常活跃小众

Risk Assessment

风险评估

  • [风险1]:[缓解方案]
  • [风险2]:[缓解方案]

Recommendations

建议

  1. Primary recommendation: [Option] because [reasons]
  2. Alternative: [Option] if [conditions]
  1. 核心建议:[方案],因为[原因]
  2. 替代方案:[方案],若[条件]

Implementation Roadmap

实施路线图

  1. Step 1
  2. Step 2
  3. Step 3
  1. 步骤1
  2. 步骤2
  3. 步骤3

References

参考文献

  • Source 1
  • Source 2
undefined
  • 来源1
  • 来源2
undefined

Research Tips

研究技巧

Effective Web Searches

高效网络搜索

  • Use specific technical terms
  • Include version numbers when relevant
  • Search for "[technology] vs [alternative]"
  • Look for "[technology] production experience"
  • 使用具体技术术语
  • 必要时包含版本号
  • 搜索“[技术] vs [替代方案]”
  • 查找“[技术] 生产实战经验”

Evaluating Sources

来源评估

  • Prefer official documentation
  • Check article/post dates
  • Look for hands-on experience reports
  • Verify claims with multiple sources
  • 优先选择官方文档
  • 检查文章/发布日期
  • 寻找实战经验报告
  • 多源验证结论

Context7 Usage

Context7 使用方法

  • Resolve library ID first:
    mcp__context7__resolve-library-id
  • Query with specific questions:
    mcp__context7__query-docs
  • 先解析库ID:
    mcp__context7__resolve-library-id
  • 用具体问题查询:
    mcp__context7__query-docs

Present Results to User

向用户呈现结果

When delivering research:
  • Start with executive summary
  • Provide clear recommendations
  • Include comparative tables
  • List sources for verification
  • Acknowledge limitations
交付研究成果时:
  • 以执行摘要开头
  • 提供清晰建议
  • 包含对比表格
  • 列出用于验证的来源
  • 说明局限性

Troubleshooting

故障排除

"Conflicting information found"
  • Note the discrepancy in report
  • Check source dates (newer may be more accurate)
  • Look for official clarification
  • Present both perspectives if unresolved
"Insufficient information"
  • Expand search terms
  • Try different source types
  • Acknowledge gaps in report
  • Suggest ways to gather more data
“发现冲突信息”
  • 在报告中注明差异
  • 检查来源日期(较新的信息可能更准确)
  • 寻找官方澄清
  • 若未解决则同时呈现两种观点
“信息不足”
  • 扩展搜索关键词
  • 尝试不同类型的来源
  • 在报告中说明信息缺口
  • 建议补充数据的方法