/analyze-and-generate-skills

Analyze the codebase to detect the tech stack and generate specialized skills for optimal code generation.

Process Overview

This command orchestrates 10 specialized agents to comprehensively analyze your codebase and generate tailored skills.

Step 1: Tech Stack Detection

First, activate the tech-stack-detector agent to identify all technologies in use:

• Scan dependency files (package.json, requirements.txt, Gemfile, Cargo.toml, go.mod, etc.)
• Identify frontend frameworks, backend frameworks, databases, testing tools, and build systems
• Create a comprehensive tech stack inventory

Step 2: Parallel Agent Analysis

Based on the detected tech stack, activate the following agents in parallel:

1. security-analyzer - Always runs to generate security best practices
2. performance-analyzer - Always runs to generate performance optimization guidelines
3. react-analyzer - Only if React is detected
4. backend-analyzer - Only if a backend framework is detected
5. frontend-analyzer - Runs if any frontend framework is detected
6. database-analyzer - Only if a database system is detected
7. testing-analyzer - Runs if testing frameworks are detected
8. architecture-analyzer - Always runs to analyze code structure
9. dependency-analyzer - Always runs to analyze dependencies
10. code-quality-analyzer - Always runs to establish quality standards

Step 3: Skill Generation

Each agent generates a skill in the .claude/skills/ directory following this structure:

• Each skill is created in its own subdirectory (prefixed with "agentic-jumpstart-")
• Inside each subdirectory, there is a SKILL.md file (case-sensitive, uppercase) that describes how to apply the skill
• The SKILL.md file must follow the official Skill format with YAML frontmatter and Markdown instructions

SKILL.md File Format:

Each SKILL.md file must have:

1. YAML frontmatter (between --- markers) starting on line 1 with no blank lines before it
2. Required fields in the frontmatter:
   • name: The skill name (e.g., "agentic-jumpstart-security")
   • description: A clear description that Claude uses to decide when to apply the skill. This is critical - include specific capabilities and trigger terms users would mention.
3. Markdown instructions after the frontmatter that describe how to apply the skill

Example SKILL.md structure:

---
name: agentic-jumpstart-security
description: Security best practices and guidelines specific to your tech stack. Use when writing secure code, reviewing security concerns, or when the user mentions security, vulnerabilities, or authentication.
---

# Security Best Practices

[Markdown instructions on how to apply security practices...]

Generated skills follow this directory structure (all prefixed with "agentic-jumpstart-"):

• .claude/skills/agentic-jumpstart-security/ → SKILL.md - Security guidelines specific to your stack
• .claude/skills/agentic-jumpstart-performance/ → SKILL.md - Performance patterns and optimizations
• .claude/skills/agentic-jumpstart-react/ → SKILL.md - React-specific patterns (if React detected)
• .claude/skills/agentic-jumpstart-backend/ → SKILL.md - Backend framework patterns (if backend detected)
• .claude/skills/agentic-jumpstart-frontend/ → SKILL.md - Frontend framework patterns (if frontend detected)
• .claude/skills/agentic-jumpstart-database/ → SKILL.md - Database and ORM patterns (if database detected)
• .claude/skills/agentic-jumpstart-testing/ → SKILL.md - Testing strategies and patterns
• .claude/skills/agentic-jumpstart-architecture/ → SKILL.md - Codebase structure and organization patterns
• .claude/skills/agentic-jumpstart-dependency-management/ → SKILL.md - Dependency management best practices
• .claude/skills/agentic-jumpstart-code-quality/ → SKILL.md - Linting, formatting, and quality standards

Execution Flow

1. Initialize: Create .claude/skills/ directory if it doesn't exist
2. Detect Stack: Run tech-stack-detector agent and capture results
3. Activate Agents: Based on detection results, activate relevant agents:
   • Always: security, performance, architecture, dependency, code-quality
   • Conditionally: react (if React), backend (if backend), frontend (if frontend), database (if database), testing (if testing frameworks)
4. Generate Skills: Each agent creates a subdirectory in .claude/skills/ prefixed with "agentic-jumpstart-" and generates a SKILL.md file (case-sensitive, uppercase) inside it. The SKILL.md file must include:
   • YAML frontmatter (between --- markers) starting on line 1 with no blank lines before it
   • Required name field matching the subdirectory name
   • Required description field with specific capabilities and trigger terms that users would naturally say
   • Markdown instructions after the frontmatter with stack-specific guidance on how to apply the skill
5. Summary: Provide a summary of generated skills and how to use them

Usage

Run /analyze-and-generate-skills in your workspace. The command will:

• Scan your codebase comprehensively
• Detect your complete tech stack
• Generate 5-10 specialized skills based on what's detected
• Create subdirectories in .claude/skills/ prefixed with "agentic-jumpstart-" for each skill
• Generate a SKILL.md file (case-sensitive, uppercase) in each subdirectory with:
  • YAML frontmatter containing name and description fields
  • A clear description that includes specific capabilities and trigger terms
  • Markdown instructions describing how to apply that skill
• Save all skills for automatic future reference

Output

The command will create skill directories and files following this structure:

Skill Directory Structure:

.claude/skills/
├── agentic-jumpstart-security/
│   └── SKILL.md          # Always generated
├── agentic-jumpstart-performance/
│   └── SKILL.md          # Always generated
├── agentic-jumpstart-react/
│   └── SKILL.md          # Generated if React detected
├── agentic-jumpstart-backend/
│   └── SKILL.md          # Generated if backend framework detected
├── agentic-jumpstart-frontend/
│   └── SKILL.md          # Generated if frontend framework detected
├── agentic-jumpstart-database/
│   └── SKILL.md          # Generated if database detected
├── agentic-jumpstart-testing/
│   └── SKILL.md          # Generated if testing frameworks detected
├── agentic-jumpstart-architecture/
│   └── SKILL.md          # Always generated
├── agentic-jumpstart-dependency-management/
│   └── SKILL.md          # Always generated
└── agentic-jumpstart-code-quality/
    └── SKILL.md          # Always generated

Each SKILL.md file follows the official Skill format:

Required Structure:

1. YAML Frontmatter (between --- markers):

   • Must start on line 1 with no blank lines before it
   • Must include name field (matching the subdirectory name)
   • Must include description field (critical - Claude uses this to decide when to apply the skill)
   • The description should answer: "What does this Skill do?" and "When should Claude use it?" with specific trigger terms

2. Markdown Instructions (after the frontmatter):

   • Best practices specific to your tech stack
   • Code patterns and conventions
   • Implementation guidelines
   • Examples relevant to your codebase

Important Notes:

• The filename must be exactly SKILL.md (uppercase, case-sensitive)
• The description field is crucial - Claude reads descriptions to find relevant Skills, so include keywords users would naturally say
• Skills are model-invoked: Claude automatically applies relevant Skills when your request matches their description
• These skills will be automatically referenced in future code generation prompts to ensure the highest quality output that matches your codebase's patterns, conventions, and best practices

Agent Coordination

The command coordinates all agents by:

1. Sharing tech stack detection results with all agents
2. Ensuring conditional agents only run when relevant technologies are detected
3. Collecting all generated skill content
4. Creating subdirectories in .claude/skills/ prefixed with "agentic-jumpstart-" for each skill
5. Writing SKILL.md files (case-sensitive, uppercase) in each subdirectory with:
   • YAML frontmatter (between --- markers) starting on line 1
   • Required name and description fields in the frontmatter
   • A clear description with specific capabilities and trigger terms
   • Markdown instructions after the frontmatter on how to apply the skill
6. Providing a comprehensive summary of what was generated