>_

router-engineering

Master orchestration for routing technical problems through 30+ engineering skills - AI/ML, software development, data, APIs, and Claude Code framework

$ Instalar

git clone https://github.com/vasilyu1983/AI-Agents-public /tmp/AI-Agents-public && cp -r /tmp/AI-Agents-public/frameworks/claude-code-kit/framework/skills/router-engineering ~/.claude/skills/AI-Agents-public

// tip: Run this command in your terminal to install the skill

SKILL.md
View on GitHub →

name: router-engineering description: Master orchestration for routing technical problems through 30+ engineering skills - AI/ML, software development, data, APIs, and Claude Code framework version: "1.0"

Router: Engineering

Master orchestrator that routes technical problems, implementation questions, and development tasks through the complete engineering skill set.

Decision Tree: Where to Start?

TECHNICAL QUESTION
    │
    ├─► "Build an AI agent" ────────────► ai-agents
    │                                      └─► MCP, guardrails, orchestration
    │
    ├─► "Fine-tune / train LLM" ────────► ai-llm
    │                                      └─► PEFT, LoRA, evaluation
    │
    ├─► "Optimize LLM inference" ───────► ai-llm-inference
    │                                      └─► vLLM, quantization, serving
    │
    ├─► "Build RAG system" ─────────────► ai-rag
    │                                      └─► Chunking, retrieval, reranking
    │
    ├─► "ML/Data science project" ──────► ai-ml-data-science
    │                                      └─► EDA, modeling, evaluation
    │
    ├─► "Time series forecasting" ──────► ai-ml-timeseries
    │                                      └─► LightGBM, Transformers, validation
    │
    ├─► "MLOps / ML security" ──────────► ai-mlops
    │                                      └─► Deployment, drift, security
    │
    ├─► "Write prompts for Claude" ─────► ai-prompt-engineering
    │                                      └─► Templates, validation, patterns
    │
    ├─► "Build backend API" ────────────► software-backend
    │                                      └─► Node.js, Python, Go, Prisma
    │
    ├─► "Build frontend app" ───────────► software-frontend
    │                                      └─► Next.js, React, Tailwind
    │
    ├─► "Build mobile app" ─────────────► software-mobile
    │                                      └─► Swift, Kotlin, React Native
    │
    ├─► "Design system architecture" ───► software-architecture-design
    │                                      └─► Microservices, CQRS, scaling
    │
    ├─► "Web3 / blockchain" ────────────► software-crypto-web3
    │                                      └─► Solidity, Rust, smart contracts
    │
    ├─► "Security / OWASP" ─────────────► software-security-appsec
    │                                      └─► Auth, input validation, crypto
    │
    ├─► "Design REST/GraphQL API" ──────► dev-api-design
    │                                      └─► OpenAPI, versioning, errors
    │
    ├─► "Manage dependencies" ──────────► dev-dependency-management
    │                                      └─► npm, pip, cargo, security
    │
    ├─► "Plan implementation" ──────────► dev-workflow-planning
    │                                      └─► /brainstorm, /write-plan
    │
    ├─► "Build data pipeline" ──────────► data-lake-platform
    │                                      └─► dlt, SQLMesh, Iceberg, DuckDB
    │
    ├─► "Optimize SQL queries" ─────────► data-sql-optimization
    │                                      └─► EXPLAIN, indexing, tuning
    │
    ├─► "Create Claude Code agent" ─────► claude-code-agents
    │                                      └─► YAML frontmatter, tools
    │
    ├─► "Create slash command" ─────────► claude-code-commands
    │                                      └─► $ARGUMENTS, templates
    │
    ├─► "Create hook automation" ───────► claude-code-hooks
    │                                      └─► PreToolUse, PostToolUse, Stop
    │
    ├─► "Setup MCP server" ─────────────► claude-code-mcp
    │                                      └─► Database, filesystem, APIs
    │
    ├─► "Configure CLAUDE.md" ──────────► claude-code-project-memory
    │                                      └─► Project context, standards
    │
    ├─► "Create skill" ─────────────────► claude-code-skills
    │                                      └─► Progressive disclosure, resources/
    │
    └─► "Full technical design" ────────► COMPREHENSIVE ANALYSIS
                                           └─► Architecture + implementation plan

Domain Detection

Domain 1: AI/ML ENGINEERING

Triggers: "AI", "ML", "LLM", "agent", "model", "train", "fine-tune", "RAG", "embedding", "inference", "prompt"

Primary Skills:

SkillWhen to Use
ai-agentsMulti-agent systems, MCP, orchestration
ai-llmFine-tuning, PEFT, LoRA, evaluation
ai-llm-inferencevLLM, quantization, serving optimization
ai-ragRetrieval-augmented generation, search
ai-ml-data-scienceEDA, feature engineering, modeling
ai-ml-timeseriesForecasting, temporal validation
ai-mlopsDeployment, monitoring, security
ai-prompt-engineeringPrompt design, templates

Skill Chain - Build AI Agent:

Requirements → ai-agents (architecture) → ai-rag (if retrieval needed)
    → ai-prompt-engineering (prompts) → ai-mlops (deployment)

Domain 2: SOFTWARE DEVELOPMENT

Triggers: "build", "implement", "code", "frontend", "backend", "mobile", "API", "database"

Primary Skills:

SkillWhen to Use
software-backendREST APIs, databases, auth, Node/Python/Go
software-frontendNext.js, React, TypeScript, Tailwind
software-mobileiOS (Swift), Android (Kotlin), React Native
software-architecture-designSystem design, microservices, scaling
software-crypto-web3Blockchain, smart contracts, DeFi
software-security-appsecOWASP, auth, input validation
software-code-reviewReview patterns, checklists
software-ui-ux-designUI patterns, accessibility
software-ux-researchUser research, gap analysis

Skill Chain - Full-Stack App:

Requirements → software-architecture-design → software-backend (API)
    → software-frontend (UI) → software-security-appsec (security audit)

Domain 3: DATA ENGINEERING

Triggers: "data pipeline", "ETL", "lakehouse", "SQL", "query optimization", "data warehouse"

Primary Skills:

SkillWhen to Use
data-lake-platformPipelines, Iceberg, DuckDB, dlt
data-sql-optimizationQuery tuning, indexes, EXPLAIN

Skill Chain - Data Platform:

Requirements → data-lake-platform (ingestion + storage)
    → data-sql-optimization (query layer) → ai-ml-data-science (analytics)

Domain 4: API & INTEGRATION

Triggers: "REST", "GraphQL", "gRPC", "OpenAPI", "API design", "versioning"

Primary Skills:

SkillWhen to Use
dev-api-designAPI design, OpenAPI, versioning
software-backendImplementation
dev-dependency-managementPackage management, security

Skill Chain - API Development:

dev-api-design (spec) → software-backend (implement)
    → dev-dependency-management (dependencies) → qa-testing-strategy (tests)

Domain 5: CLAUDE CODE FRAMEWORK

Triggers: "Claude Code", "agent", "skill", "command", "hook", "MCP", "CLAUDE.md"

Primary Skills:

SkillWhen to Use
claude-code-agentsCreate agents with frontmatter
claude-code-commandsSlash commands with $ARGUMENTS
claude-code-hooksEvent automation (Pre/PostToolUse)
claude-code-mcpExternal data connections
claude-code-project-memoryCLAUDE.md configuration
claude-code-skillsProgressive disclosure skills

Skill Chain - Full Claude Code Setup:

claude-code-project-memory (CLAUDE.md) → claude-code-skills (knowledge)
    → claude-code-agents (subagents) → claude-code-commands (shortcuts)
    → claude-code-hooks (automation) → claude-code-mcp (external data)

Skill Registry

AI/ML Skills (9)

SkillPurposeKey Outputs
ai-agentsMulti-agent systemsArchitecture, MCP config, guardrails
ai-llmLLM developmentFine-tuning config, evaluation
ai-llm-inferenceInference optimizationvLLM config, quantization
ai-ragRAG systemsChunking strategy, retrieval pipeline
ai-ml-data-scienceData scienceEDA, models, evaluation
ai-ml-timeseriesTime seriesForecasting models, validation
ai-mlopsML operationsDeployment, monitoring
ai-prompt-engineeringPrompt designTemplates, validation
agent-fleet-operationsAgent servicesFleet management, economics

Software Skills (9)

SkillPurposeKey Outputs
software-backendBackend developmentAPIs, database, auth
software-frontendFrontend developmentComponents, state, routing
software-mobileMobile developmentiOS, Android, cross-platform
software-architecture-designSystem designArchitecture diagrams, decisions
software-crypto-web3BlockchainSmart contracts, DeFi
software-security-appsecSecurityOWASP compliance, auth
software-code-reviewCode reviewReview checklists
software-ui-ux-designUI/UX designDesign systems, accessibility
software-ux-researchUX researchUser research, benchmarks

Data Skills (2)

SkillPurposeKey Outputs
data-lake-platformData pipelinesIngestion, transformation, storage
data-sql-optimizationSQL tuningQuery optimization, indexes

Development Workflow (3)

SkillPurposeKey Outputs
dev-api-designAPI designOpenAPI specs, contracts
dev-dependency-managementDependenciesLockfiles, security
dev-workflow-planningPlanningImplementation plans

Claude Code Framework (6)

SkillPurposeKey Outputs
claude-code-agentsAgent creationAgent YAML, tools config
claude-code-commandsCommandsSlash command templates
claude-code-hooksHooksEvent automation scripts
claude-code-mcpMCP serversDatabase/API connections
claude-code-project-memoryCLAUDE.mdProject configuration
claude-code-skillsSkillsProgressive disclosure docs

Routing Logic

Keyword-Based Routing

KEYWORDS -> SKILL MAPPING

"agent", "multi-agent", "MCP", "guardrails" -> ai-agents
"fine-tune", "LoRA", "PEFT", "train model" -> ai-llm
"vLLM", "inference", "quantization", "serving" -> ai-llm-inference
"RAG", "retrieval", "embedding", "vector" -> ai-rag
"EDA", "feature engineering", "modeling" -> ai-ml-data-science
"time series", "forecast", "temporal" -> ai-ml-timeseries
"MLOps", "model deployment", "drift" -> ai-mlops
"prompt", "template", "instruction" -> ai-prompt-engineering

"backend", "API", "Node.js", "FastAPI", "Prisma" -> software-backend
"frontend", "Next.js", "React", "Tailwind" -> software-frontend
"mobile", "iOS", "Android", "Swift", "Kotlin" -> software-mobile
"architecture", "microservices", "CQRS" -> software-architecture-design
"blockchain", "Solidity", "Web3", "smart contract" -> software-crypto-web3
"security", "OWASP", "auth", "vulnerability" -> software-security-appsec
"code review", "review checklist" -> software-code-review
"UI", "design system", "accessibility" -> software-ui-ux-design
"UX research", "user research", "JTBD" -> software-ux-research

"data pipeline", "ETL", "lakehouse", "dlt" -> data-lake-platform
"SQL optimization", "query tuning", "EXPLAIN" -> data-sql-optimization

"REST API", "GraphQL", "OpenAPI", "gRPC" -> dev-api-design
"npm", "pip", "cargo", "dependencies" -> dev-dependency-management
"plan", "brainstorm", "implementation" -> dev-workflow-planning

"Claude Code agent", "subagent" -> claude-code-agents
"slash command", "$ARGUMENTS" -> claude-code-commands
"hook", "PreToolUse", "PostToolUse" -> claude-code-hooks
"MCP server", "database connection" -> claude-code-mcp
"CLAUDE.md", "project memory" -> claude-code-project-memory
"skill", "progressive disclosure" -> claude-code-skills

Context-Based Routing

User ContextPrimary SkillSupporting Skills
Building AI appai-agentsai-rag, ai-prompt-engineering
Optimizing LLMai-llm-inferenceai-mlops
Full-stack appsoftware-architecture-designsoftware-backend, software-frontend
Mobile appsoftware-mobilesoftware-backend, software-ui-ux-design
Data platformdata-lake-platformdata-sql-optimization, ai-ml-data-science
API developmentdev-api-designsoftware-backend, software-security-appsec
Claude Code setupclaude-code-project-memoryAll claude-code-* skills

Skill Chain Patterns

Pattern 1: AI Agent Development

START
  │
  ▼
ai-agents ──────────────────► Agent Architecture
  │
  ▼
ai-rag (if retrieval) ──────► Retrieval Pipeline
  │
  ▼
ai-prompt-engineering ──────► Prompt Templates
  │
  ▼
ai-mlops ───────────────────► Deployment Config
  │
  ▼
agent-fleet-operations ─────► Service Design
  │
  ▼
PRODUCTION AGENT

Pattern 2: Full-Stack Application

START
  │
  ▼
software-architecture-design ► System Design
  │
  ├──────────────────────────────────────────┐
  ▼                                          ▼
software-backend ──► API         software-frontend ──► UI
  │                                          │
  └────────────────────┬─────────────────────┘
                       ▼
         software-security-appsec ──► Security Audit
                       │
                       ▼
                  DEPLOYED APP

Pattern 3: Data Platform

START
  │
  ▼
data-lake-platform ─────────► Ingestion + Storage
  │
  ▼
data-sql-optimization ──────► Query Layer
  │
  ▼
ai-ml-data-science ─────────► Analytics
  │
  ▼
software-frontend ──────────► Dashboard
  │
  ▼
DATA PLATFORM LIVE

Pattern 4: Claude Code Framework Setup

START
  │
  ▼
claude-code-project-memory ─► CLAUDE.md
  │
  ▼
claude-code-skills ─────────► Domain Knowledge
  │
  ▼
claude-code-agents ─────────► Specialized Agents
  │
  ▼
claude-code-commands ───────► User Shortcuts
  │
  ▼
claude-code-hooks ──────────► Automation
  │
  ▼
claude-code-mcp ────────────► External Data
  │
  ▼
COMPLETE CLAUDE CODE SETUP

Comprehensive Analysis Mode

For full technical analysis, invoke skills in parallel:

Layer 1: Requirements & Design (Parallel)

SkillOutputPurpose
software-architecture-designSystem designArchitecture decisions
dev-api-designAPI contractsInterface design
software-ui-ux-designUI patternsUser interface

Layer 2: Implementation (Based on Stack)

StackPrimary Skills
AI/MLai-agents, ai-rag, ai-llm
Full-stacksoftware-backend, software-frontend
Mobilesoftware-mobile, software-backend
Datadata-lake-platform, data-sql-optimization

Layer 3: Quality & Security

SkillPurpose
software-security-appsecSecurity audit
software-code-reviewCode quality
Route to → router-operationsTesting, deployment

Cross-Router Handoffs

To router-startup

When user shifts to business concerns:

  • "How should I price this?" → startup-business-models
  • "Who are competitors?" → startup-competitive-analysis
  • "Go to market strategy?" → startup-go-to-market

To router-operations

When user shifts to operations:

  • "How do I test this?" → qa-testing-strategy
  • "Deploy to production" → ops-devops-platform
  • "Monitor performance" → qa-observability

From router-startup

When router-startup detects technical needs:

  • "Build the MVP" → Route here for implementation
  • "Technical architecture" → Route here for design

Output Templates

Quick Analysis Output

## Technical Analysis: {{TOPIC}}

**Domain Detected**: {{DOMAIN}}
**Primary Skill**: {{SKILL}}
**Supporting Skills**: {{LIST}}

### Recommended Approach
1. {{STEP_1}} - Use {{SKILL}}
2. {{STEP_2}} - Use {{SKILL}}
3. {{STEP_3}} - Use {{SKILL}}

### Key Considerations
- {{CONSIDERATION_1}}
- {{CONSIDERATION_2}}

### Skills to Invoke
- {{SKILL_1}}: {{WHY}}
- {{SKILL_2}}: {{WHY}}

Resources

ResourcePurpose
routing-logic.mdDetailed routing rules
skill-chain-patterns.mdImplementation patterns
technology-selection.mdStack recommendations

Templates

TemplatePurpose
technical-analysis-report.mdFull analysis
architecture-decision.mdADR template
implementation-plan.mdImplementation planning

Data

FilePurpose
skill-registry.jsonEngineering skills index
sources.jsonReference sources

Repository

vasilyu1983
vasilyu1983
Author
vasilyu1983/AI-Agents-public/frameworks/claude-code-kit/framework/skills/router-engineering
21
Stars
6
Forks
Updated4d ago
Added1w ago

Actions

View on GitHubDownload ZIPReport Issue

Related Skills

skill-writer
pytorch/pytorch
95.4k
docstring
pytorch/pytorch
95.4k
at-dispatch-v2
pytorch/pytorch
95.4k
add-uint-support
pytorch/pytorch
95.4k
skill-creator
openai/codex
54.7k