最佳实践调研
ln-002-best-practices-researcher
by levnikolaevich
结合MCP Ref、Context7和WebSearch快速查最佳实践、官方文档与版本信息,并生成guide、manual、ADR或research文档,适合给项目补齐模式说明、API手册和架构决策记录。
安装
claude skill add --url github.com/levnikolaevich/claude-code-skills/tree/master/ln-002-best-practices-researcher文档
Paths: File paths (
shared/,references/,../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root.
Best Practices Researcher
Research industry standards and create project documentation in one workflow.
Purpose & Scope
- Research via MCP Ref + Context7 for standards, patterns, versions
- Create 4 types of documents from research results:
- Guide: Pattern documentation (Do/Don't/When table)
- Manual: API reference (methods/params/doc links)
- ADR: Architecture Decision Record (Q&A dialog)
- Research: Investigation document answering specific question
- Return document path for linking in Stories/Tasks
Phase 0: Stack Detection
Objective: Identify project stack to filter research queries and adapt output.
Detection:
| Indicator | Stack | Query Prefix | Official Docs |
|---|---|---|---|
*.csproj, *.sln | .NET | "C# ASP.NET Core" | Microsoft docs |
package.json + tsconfig.json | Node.js | "TypeScript Node.js" | MDN, npm docs |
requirements.txt, pyproject.toml | Python | "Python" | Python docs, PyPI |
go.mod | Go | "Go Golang" | Go docs |
Cargo.toml | Rust | "Rust" | Rust docs |
build.gradle, pom.xml | Java | "Java" | Oracle docs, Maven |
Usage:
- Add
query_prefixto all MCP search queries - Link to stack-appropriate official docs
When to Use
- ln-310-story-validator detects missing documentation
- Need to document a pattern, library, or decision
- Replaces: ln-321-guide-creator, ln-322-adr-creator, ln-323-manual-creator
Input Parameters
| Parameter | Required | Description |
|---|---|---|
| doc_type | Yes | "guide" / "manual" / "adr" / "research" |
| topic | Yes | What to document (pattern name, package name, decision title, research question) |
| story_context | No | Story/Task context for relevance |
Research Tools
| Tool | Use Case | Query Pattern |
|---|---|---|
ref_search_documentation | Standards, patterns, RFCs | "[topic] RFC standard best practices {current_year}" |
context7__resolve-library-id | Get library ID for docs | libraryName="[topic]" |
context7__query-docs | Library API, methods | topic="[stack_prefix] [topic]" |
WebSearch | Market, competitors, versions | "[topic] latest {current_year}" or "[topic] vs alternatives" |
Time-box: 5-10 minutes for research; skip if topic is trivial
Research Methodology by Type (for doc_type="research")
| Type | Focus | Primary Sources | Key Questions |
|---|---|---|---|
| Technical | Solution comparison | Docs, benchmarks, RFCs | "Which solution fits our use-case?" |
| Market | Industry landscape | Reports, blogs, articles | "What's the market size/trend?" |
| Competitor | How others solve it | Product pages, reviews, demos | "What features do competitors offer?" |
| Requirements | User needs | Feedback, support tickets, forums | "What do customers complain about?" |
| Feasibility | Can we build it? | PoC, prototypes, local tests | "Is it technically possible?" |
| Feature Demand | Feature viability | Competitor features + blogs/socials + user complaints | "Is this feature worth building?" |
Workflow
| doc_type | Purpose | Research Source | Template | Output Path | Words |
|---|---|---|---|---|---|
| guide | Pattern with Do/Don't/When table | ref_search (best practices) | guide_template.md | guides/NN-[slug].md | 300-500 |
| manual | API/library reference | context7__query-docs | manual_template.md | manuals/[pkg]-[ver].md | 300-500 |
| adr | Architecture decision | Dialog (5 questions) | adr_template.md | adrs/adr-NNN-[slug].md | 300-500 |
| research | Investigation answering question | See Methodology table above | research_template.md | research/rsh-NNN-[slug].md | 300-700 |
Common Workflow: Detect number (if needed) → Research → Generate from template → Validate (SCOPE, POSIX) → Save → Return path
Extract & Sections by doc_type:
- guide: Extract principle, 2-3 do/don'ts, sources → Sections: Principle, Our Implementation, Patterns table, Sources, Related
- manual: Extract methods, params (type/required/default), returns → Sections: Package info, Overview, Methods table, Config table, Limitations
- adr: Dialog answers → Sections: Context, Decision, Rationale, Alternatives table, Consequences, Related
- research: Findings by methodology → Sections: Question, Context, Methodology, Findings (tables!), Conclusions, Next Steps, Sources
Validation specifics: guide: patterns table present; manual: version in filename; adr: ISO date, status field; all: sources ≤ 1 year old
ADR Dialog (5 questions): Q1: Title? → Q2: Category (Strategic/Technical)? → Q3: Context? → Q4: Decision + Rationale? → Q5: Alternatives (2 with pros/cons)?
Output: File path for linking in Stories/Tasks; for ADR remind to reference in architecture.md; for Research suggest ADR if decision needed
Critical Rules
MANDATORY FILE CREATION:
- ALL research MUST end with file creation (guide/manual/ADR/research document)
- Create target directory if missing (docs/guides/, docs/manuals/, docs/adrs/, docs/research/)
- No exceptions — file creation is required for ALL invocations
NO_CODE_EXAMPLES (ALL document types):
| Forbidden | Allowed |
|---|---|
| Code snippets | Tables (params, config, alternatives) |
| Implementation examples | ASCII diagrams, Mermaid diagrams |
| Code blocks >1 line | Method signatures (1 line inline) |
| Links to official docs |
Format Priority (STRICT):
┌───────────────────────────────────────────────┐
│ 1. TABLES + ASCII diagrams ←── PRIORITY │
│ Params, Config, Alternatives, Flows │
├───────────────────────────────────────────────┤
│ 2. LISTS (enumerations only) │
│ Enumeration items, file lists, tools │
├───────────────────────────────────────────────┤
│ 3. TEXT (last resort) │
│ Only if cannot express as table │
└───────────────────────────────────────────────┘
| Content Type | Format |
|---|---|
| Parameters | Table: Name | Type | Required | Default |
| Configuration | Table: Option | Type | Default | Description |
| Alternatives | Table: Alt | Pros | Cons | Why Rejected |
| Patterns | Table: Do | Don't | When |
| Workflow | ASCII diagram: A → B → C |
Other Rules:
- Research ONCE per invocation; reuse results
- Cite sources with versions/dates (≤ 1 year old)
- One pattern per guide; one decision per ADR; one package per manual
- Preserve language (EN/RU) from story_context
- Link to stack-appropriate docs (Microsoft for .NET, MDN for JS, etc.)
- MANDATORY: Create target directory if missing (docs/guides/, docs/manuals/, docs/adrs/, docs/research/); file creation is required
Definition of Done
- Research completed (standards/patterns/versions extracted) - for guide/manual
- Dialog completed (5 questions answered) - for ADR
- Document generated with all required sections; no placeholders
- Standards validated (SCOPE, maintenance, POSIX)
- File saved to correct directory with proper naming
- Path returned; README updated if placeholder present
Reference Files
- Guide template:
shared/templates/guide_template.md - Manual template:
shared/templates/manual_template.md - ADR template:
shared/templates/adr_template.md - Research template:
shared/templates/research_template.md - Standards:
docs/DOCUMENTATION_STANDARDS.md(if exists) - MANDATORY READ:
shared/references/research_tool_fallback.md
Version: 3.0.1 Last Updated: 2026-02-14
相关 Skills
股票投研框架
by alondai
股票投研9点分析框架,覆盖基本面/财务/竞品/估值/宏观/情绪等维度
SEO审计工具
by amdf01-debug
智能浏览器
by ayakolin
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured commands. (Chinese localized version)
相关 MCP 服务
网页抓取
编辑精选by Anthropic
Fetch 是 MCP 官方参考服务器,让 AI 能抓取网页并转为 Markdown 格式。
✎ 这个服务器解决了 AI 直接处理网页内容时格式混乱的问题,适合需要让 Claude 分析在线文档或新闻的开发者。不过作为参考实现,它缺乏生产级的安全配置,你得自己处理反爬虫和隐私风险。
Brave 搜索
编辑精选by Anthropic
Brave Search 是让 Claude 直接调用 Brave 搜索 API 获取实时网络信息的 MCP 服务器。
✎ 如果你想让 AI 助手帮你搜索最新资讯或技术文档,这个工具能绕过传统搜索的限制,直接返回结构化数据。特别适合需要实时信息的开发者,比如查 API 更新或竞品动态。不过它依赖 Brave 的 API 配额,高频使用可能受限。
Puppeteer 浏览器控制
编辑精选by Anthropic
Puppeteer 是让 Claude 自动操作浏览器进行网页抓取和测试的 MCP 服务器。
✎ 这个服务器解决了手动编写 Puppeteer 脚本的繁琐问题,适合需要自动化网页交互的开发者,比如抓取动态内容或做端到端测试。不过,作为参考实现,它可能缺少生产级的安全防护,建议在可控环境中使用。