ln-001-standards-researcher（ln 001 标准研究员）

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.

Standards Researcher (Worker)

This skill researches industry standards and architectural patterns using MCP Ref to generate Standards Research for Story Technical Notes.

Purpose

Research industry standards, RFCs, and architectural patterns for a given Epic/Story domain. Produce a Standards Research section (tables + links, no code) for insertion into Story Technical Notes.

When to Use This Skill

This skill should be used when:

• Need to research standards and patterns BEFORE Story generation (ensures tasks follow industry best practices)
• Epic Technical Notes mention specific standards requiring documentation (OAuth, OpenAPI, WebSocket)
• Prevent situations where tasks use outdated patterns or violate RFC compliance
• Reusable for ANY skill requiring standards research

Workflow

The skill follows a 7-phase workflow focused on standards and architectural patterns.

Stack Detection → Identify → Ref Research → Existing Guides → Generate Research → Save to File → Return

Phase 0: Stack Detection

Objective: Determine project stack BEFORE research to filter queries.

Detection:

Process:

1. Check context_store.TECH_STACK if provided → use directly
2. Else: Glob for indicator files in project root
3. Store detected_stack.query_prefix for Phases 2-3

Output: detected_stack = {language, framework, query_prefix}

Skip conditions: If no stack detected → proceed without prefix (generic queries)

Phase 1: Identify Libraries

Objective: Parse Epic/Story for libraries and technology keywords.

Process:

1. Read Epic/Story description (provided as input)

- Parse Epic Technical Notes for mentioned libraries/frameworks - Parse Epic Scope In for technology keywords (authentication, rate limiting, payments, etc.) - Identify Story domain from Epic goal statement (e.g., "Add rate limiting" → domain = "rate limiting")

1. Extract library list:

- Primary libraries (explicitly mentioned) - Inferred libraries (e.g., "REST API" → FastAPI, "caching" → Redis) - Filter out well-known libraries with stable APIs (e.g., requests, urllib3)

1. Determine Story domain:

- Extract from Epic goal or Story title - Examples: rate limiting, authentication, payment processing, file upload

Output: Library list (3-5 libraries max) + Story domain

Skip conditions:

• NO libraries mentioned in Epic → Output empty Research Summary
• Trivial CRUD operation with well-known libraries → Output empty Research Summary
• Epic explicitly states "research not needed" → Skip

Phase 2: MCP Ref Research

Objective: Get industry standards and architectural patterns.

Process:

1. Focus on standards/RFCs:

- Call mcp__Ref__ref_search_documentation(query="[detected_stack.query_prefix] [story_domain] RFC standard specification") - Example: "C# ASP.NET Core rate limiting RFC standard specification" - Extract: RFC/spec references (OAuth 2.0 RFC 6749, OpenAPI 3.0, WebSocket RFC 6455)

1. Focus on architectural patterns:

- Call mcp__Ref__ref_search_documentation(query="[detected_stack.query_prefix] [story_domain] architectural patterns best practices") - Example: "TypeScript Node.js authentication architectural patterns best practices" - Extract: Middleware, Dependency Injection, Decorator pattern

Output: Standards compliance table + Architectural patterns list

Phase 3: MCP Ref Research

Objective: Get industry standards and best practices.

Process:

1. FOR EACH library + Story domain combination:

- Call mcp__Ref__ref_search_documentation(query="[detected_stack.query_prefix] [library] [domain] best practices {current_year}") - Call mcp__Ref__ref_search_documentation(query="[detected_stack.query_prefix] [domain] industry standards RFC") - Example: "C# ASP.NET Core Polly rate limiting best practices {current_year}"

1. Extract from results (NO CODE - text/tables only):

- Industry standards (RFC/spec references: OAuth 2.0, REST API, OpenAPI 3.0) - Common patterns (do/don't descriptions, anti-patterns to avoid) - Integration approaches (middleware, dependency injection, decorators) - Security considerations (OWASP compliance, vulnerability mitigation) - Official docs URLs (link to stack-appropriate authoritative sources)

1. Store results for Research Summary compilation

Output: Standards compliance table (RFC/Standard name, how to comply) + Best practices list

Phase 4: Scan Existing Guides

Objective: Find relevant pattern guides in docs/guides/ directory.

Process:

1. Scan guides directory:

- Use Glob to find docs/guides/*.md - Read guide filenames

1. Match guides to Story domain:

- Match keywords (e.g., rate limiting guide for rate limiting Story) - Fuzzy match (e.g., "authentication" matches "auth.md", "oauth.md")

1. Collect guide paths for linking in Technical Notes

Output: Existing guides list (relative paths from project root)

Phase 5: Generate Standards Research

Objective: Compile research results into Standards Research for Story Technical Notes subsection.

NO_CODE Rule: No code snippets. Use tables + links to official docs only.

Format Priority:

┌─────────────────────────────────────┐
│ 1. TABLES + ASCII diagrams ← Priority │
│ 2. Lists (enumerations only)        │
│ 3. Text (last resort)               │
└─────────────────────────────────────┘

Output Format (Table-First):

## Standards Research

**Standards compliance:**

| Standard | Requirement | How to Comply | Reference |
|----------|-------------|---------------|-----------|
| RFC 6749 | OAuth 2.0 | Use PKCE for public clients | [RFC 6749](url) |
| RFC 6585 | Rate Limiting | Return 429 + Retry-After | [RFC 6585](url) |

**Architectural patterns:**

| Pattern | When to Use | Reference |
|---------|-------------|-----------|
| Middleware | Request interception | [Official docs](url) |
| Decorator | Cross-cutting concerns | [Official docs](url) |

**Existing guides:**
- [guide_path.md](guide_path.md) - Brief description

Output: Standards Research (Markdown string) for insertion into Story Technical Notes subsection

Important notes:

• Focus on STANDARDS and PATTERNS only (no library details - libraries researched at Task level)
• Prefer official docs and RFC standards over blog posts
• Link to stack-appropriate docs (Microsoft docs for.NET, MDN for JS, etc.)
• If Standards Research is empty (no standards/patterns) → Skip Phase 6, return "No standards research needed"

Phase 6: Save Research to File

Objective: Save Standards Research to standalone file for reusability and knowledge base.

MANDATORY: All research MUST be saved to file, even if returned as string to caller.

Process:

1. Determine next research number:

- Glob for docs/research/rsh-*.md files - Extract numbers, find max - Next number = max + 1 (or 001 if no files exist)

1. Generate filename:

- Format: rsh-{number:03d}-{slug}.md - Slug from story_domain (e.g., "rate-limiting", "oauth-authentication") - Example: rsh-042-rate-limiting.md

1. Create research document using template:

# Standards Research: {Story Domain}

**Created:** {ISO date}
**Epic:** {Epic ID if available}
**Research Type:** Standards & Architectural Patterns

## Question

What industry standards and architectural patterns apply to {story_domain}?

## Context

{Brief description from Epic/Story}

## Methodology

- **Standards:** MCP Ref search for RFCs and specifications
- **Patterns:** MCP Ref search for architectural patterns and best practices
- **Stack:** {detected_stack.language} {detected_stack.framework}

## Findings

{Insert Standards Research content from Phase 5 here}

## Conclusions

{1-2 sentences summarizing key standards/patterns that must be followed}

## Next Steps

- Reference this research in Story Technical Notes
- Link from architecture.md if patterns affect system design
- Create ADR if architectural decision needed

## Sources

{List of all MCP Ref search URLs with dates}

1. Save file:

- Write to docs/research/{filename} - If docs/research/ doesn't exist: create directory - Validate file saved successfully

1. Update README (if exists):

- Check for docs/research/README.md - If exists and has <!-- PLACEHOLDER -->: append research entry - Format: - [{filename}]({filename}) - {one-line summary}

Output: File path (e.g., docs/research/rsh-042-rate-limiting.md)

Skip conditions:

• Standards Research is empty → do not create file
• Target directory missing and cannot be created → warn user, skip file creation

Phase 7: Return Results

Return to calling skill (ln-220, ln-310):

1. Standards Research string (Markdown) for insertion into Story Technical Notes
2. File path (string) for linking: docs/research/rsh-{NNN}-{slug}.md

Format:

{
  "standards_research": "<markdown string>",
  "file_path": "docs/research/rsh-042-rate-limiting.md"
}

If calling skill expects only string, return Standards Research string only. File is created regardless.

Integration

Dependencies:

• MCP Ref (ref_search_documentation) - industry standards and patterns
• Glob (scan docs/guides/)

Input parameters:

• epic_description (string) - Epic Technical Notes + Scope In + Goal
• story_domain (string, optional) - Story domain (e.g., "rate limiting")

Output format:

• Primary: Markdown string (Standards Research for Story Technical Notes subsection)
• Secondary: File path (string): docs/research/rsh-{NNN}-{slug}.md
• Content: Standards + Patterns (libraries researched at Task level)
• Note: File is ALWAYS created (Phase 6)

Time-Box and Performance

Time-box: 15-20 minutes maximum per Epic

Performance:

• Research is done ONCE per Epic
• Results reused for all Stories (5-10 Stories benefit from single research)
• Parallel MCP calls when possible (Context7 + Ref)

Token efficiency:

• Context7: max 3000 tokens per library
• Total: ~10,000 tokens for typical Epic (3-4 libraries)

Critical Rules

• MANDATORY FILE CREATION: All research MUST be saved to docs/research/rsh-{NNN}-{slug}.md file (Phase 6); no exceptions
• NO_CODE: Output contains tables and links to official docs only; no code snippets
• Format Priority: Tables + ASCII diagrams first, lists second, text last resort
• Stack-aware queries: All MCP Ref calls must include detected query_prefix (e.g., "C# ASP.NET Core")
• Standards over libraries: Focus on RFCs and architectural patterns; library details are researched at Task level
• Time-box: Maximum 15-20 minutes per Epic; research is done once and reused for all Stories

Definition of Done

• Stack detected (or skipped if undetectable) and query_prefix set
• Libraries and Story domain extracted from Epic/Story description
• MCP Ref research completed for standards/RFCs and architectural patterns
• Existing guides in docs/guides/ scanned and matched
• Standards Research output generated in Markdown (tables + links, no code)
• Research saved to file: docs/research/rsh-{NNN}-{slug}.md created with all required sections
• README updated (if docs/research/README.md exists and has placeholder)
• Output returned to calling skill: Standards Research string + file path

Reference Files

Tools:

• mcp__Ref__ref_search_documentation() - Search best practices and standards
• Glob - Scan docs/guides/ directory

Guidelines:

• MANDATORY READ: Load references/research_guidelines.md — research quality guidelines (official docs > blog posts, prefer LTS versions)
• MANDATORY READ: Load shared/references/research_tool_fallback.md

Version: 3.1.0 Last Updated: 2026-02-14