Claude Code Integration

Integrate Skill Engine with Claude Code for powerful AI-assisted development.

Overview

Claude Code is Anthropic's official CLI tool that provides AI assistance directly in your terminal. Skill Engine integrates seamlessly with Claude Code through the Model Context Protocol (MCP).

Quick Setup

1. Install Skill Engine

curl -fsSL https://dqkbk9o7ynwhxfjx.public.blob.vercel-storage.com/install.sh | sh

2. Start MCP Server

skill serve

This starts the MCP server on stdio, which Claude Code can connect to.

3. Configure Claude Code

Add to your ~/.config/claude/mcp.json:

{
  "mcpServers": {
    "skill-engine": {
      "command": "skill",
      "args": ["serve"]
    }
  }
}

4. Verify Connection

Start Claude Code:

claude

Claude Code will automatically connect to the Skill Engine MCP server and have access to all your installed skills.

Available Tools

When connected, Claude Code can use these MCP tools:

execute

Execute any skill tool with parameters.

Example:

You: "List all Kubernetes pods"
Claude: *uses skill-engine/execute*
       skill_name: kubernetes
       tool_name: get
       parameters: {resource: "pods"}

list_skills

Discover available skills with pagination.

Example:

You: "What skills are available?"
Claude: *uses skill-engine/list_skills*

search_skills

Semantic search across all tools.

Example:

You: "Find tools for managing infrastructure"
Claude: *uses skill-engine/search_skills*
       query: "managing infrastructure"

Usage Examples

Infrastructure Management

You: "Show me all pods in the production namespace"

Claude: I'll check the Kubernetes pods in the production namespace.
        [Uses skill-engine/execute: kubernetes/get]
        
        Found 12 pods running in production:
        - api-server-abc123 (Running)
        - worker-def456 (Running)
        ...

GitHub Operations

You: "Create an issue in the my-repo repository"

Claude: I'll create an issue for you.
        [Uses skill-engine/execute: github/create-issue]
        
        Created issue #42: "Feature request"
        https://github.com/user/my-repo/issues/42

Database Queries

You: "Query the users table for active accounts"

Claude: I'll query the database.
        [Uses skill-engine/execute: postgres/query]
        
        Found 1,234 active users

Configuration

Environment Variables

Pass environment variables to skills:

{
  "mcpServers": {
    "skill-engine": {
      "command": "skill",
      "args": ["serve"],
      "env": {
        "KUBECONFIG": "/path/to/kubeconfig",
        "DATABASE_URL": "postgresql://localhost/mydb"
      }
    }
  }
}

Custom Skill Path

Specify a custom skill directory:

{
  "mcpServers": {
    "skill-engine": {
      "command": "skill",
      "args": ["serve", "--skills-dir", "/custom/path"]
    }
  }
}

Troubleshooting

MCP Server Not Connecting

Problem: Claude Code can't connect to Skill Engine

Solution:

# Test MCP server manually
skill serve

# Check if skill command is in PATH
which skill

# Verify config file exists
cat ~/.config/claude/mcp.json

Skills Not Found

Problem: Claude Code says skills are not available

Solution:

# List installed skills
skill list

# Install example skills
skill install ./examples/native-skills/kubernetes-skill

Permission Errors

Problem: Skills fail with permission denied

Solution:

# Check skill configuration
skill config kubernetes

# Verify environment variables
env | grep KUBECONFIG

Best Practices

1. Explicit Skill Names

Be specific when asking Claude to use skills:

✅ "Use the kubernetes skill to list pods"
❌ "Show me pods" (Claude might not know which skill to use)

2. Provide Context

Give Claude the information it needs:

✅ "List pods in the production namespace using kubectl"
❌ "List pods" (missing namespace context)

3. Chain Operations

Claude can execute multiple skills in sequence:

You: "Deploy the latest version and check if pods are running"

Claude: I'll deploy and verify:
        1. [skill-engine/execute: kubernetes/apply]
        2. [skill-engine/execute: kubernetes/get]
        
        Deployment successful, 3 pods running.

Advanced Usage

Custom Instructions

Add Skill Engine context to your Claude Code instructions:

# .claude/instructions.md

When managing infrastructure:
- Use the kubernetes skill for K8s operations
- Use the terraform skill for IaC changes
- Always verify changes before applying

Available skills: kubernetes, terraform, aws, github

Workflow Automation

Create workflows that Claude can execute:

You: "Run the deployment workflow"

Claude: I'll execute the deployment workflow:
        1. Run tests
        2. Build Docker image
        3. Apply Kubernetes manifests
        4. Verify deployment
        5. Update monitoring

Related Documentation

• MCP Protocol Reference
• Skill Development Guide
• API Reference
• Example Skills

Support

• GitHub Issues
• GitHub Discussions