← Back to Context Comparison

User-Level Skills

Claude Code Skills loaded from ~/.claude/skills/

~/.claude/skills/

User-level skills are loaded at session start and available across ALL projects. These skills provide general-purpose capabilities that aren't tied to specific infrastructure or projects. They represent personal preferences, cross-cutting utilities, and meta-tools for extending Claude Code itself.

12
Total Skills
2
Proactive
10
On-Demand
0
Boot Tokens

Skill Catalog

🏷️ naming-validator
Proactive - CRITICAL

Validates resource names against established naming standards BEFORE creating any infrastructure resource. Ensures consistent lowercase formatting, proper hyphenation, and correct patterns across all services.

Trigger Phrases:

  • "Install {anything}"
  • "Deploy {service}"
  • "Create a new project/service/database"
  • "Set up {component}"
references/naming-standards.md scripts/validate_name.py
🔐 secrets-security
Proactive

Secrets management, security hardening, access control, vulnerability management, and incident response. Essential for secure credential storage in Bitwarden and $HOME/projects/secrets/, security audits, and maintaining infrastructure security posture.

When to Use:

  • Storing or rotating credentials
  • Security audits and hardening
  • Incident response procedures
  • Access control configuration
references/secrets-management.md references/security-hardening.md scripts/security-audit.sh scripts/rotate-credentials.sh
🧠 openmemory
On-Demand

Store and retrieve persistent facts across Claude Code sessions using mem0 AI memory with Gemini embeddings. Enables long-term memory for preferences, decisions, patterns, and lessons learned.

When to Use:

  • Important facts emerge during conversation
  • User provides preferences to remember
  • Technical decisions that should persist
  • Patterns or workarounds discovered
API: localhost:8765 Gemini Flash + 768-dim embeddings
🦊 gitlab-integration
On-Demand

GitLab repository management, issues, merge requests, and CI/CD operations for the self-hosted gitlab.ai-servicers.com instance. Provides API commands for all GitLab operations.

Operations:

  • Repository creation and management
  • Issue and merge request workflows
  • CI/CD pipeline operations
  • Branch and user management
$HOME/projects/secrets/gitlab.env gitlab.ai-servicers.com API
🏗️ architecture-diagram-creator
On-Demand

Create comprehensive HTML architecture diagrams with SVG data flows, business objectives, features, technical architecture, and deployment documentation for any software project.

Trigger Phrases:

  • "Create an architecture diagram for {project}"
  • "Generate a high-level overview"
  • "Document the system architecture"
  • "Show me the data flow"
assets/templates/base_template.html assets/templates/architecture_components.html references/example_architecture.html
📝 claudemd-generator
On-Demand

Auto-generate CLAUDE.md project context files for services. Detects project type (Docker, Python, Node.js), scans configuration files, analyzes existing documentation, and creates appropriate templates.

Trigger Phrases:

  • "Generate CLAUDE.md"
  • "Create project context"
  • "Add documentation to this project"
  • Missing CLAUDE.md detection
Docker/Python/Node.js detection Template library
🛠️ skill-creator
On-Demand - Meta

Guide for creating effective skills that extend Claude's capabilities. Provides the skill creation process, anatomy, progressive disclosure patterns, and packaging workflow.

When to Use:

  • "Create a new skill"
  • "Update an existing skill"
  • Extending Claude Code capabilities
  • Packaging skills for distribution
scripts/init_skill.py scripts/package_skill.py
🤖 ai-communication
On-Demand

Covers all AI-to-AI communication: file sharing via MinIO aichat-files bucket and sync peer review via MCP Review Board dispatch to Gemini/Codex/Claude.

Trigger Phrases:

  • "Peer review this plan"
  • "Respond to aichat"
  • "Ask Gemini/Codex for feedback"
  • "Write to shared file"
MinIO aichat-files bucket MCP Review Board
🎨 frontend-design-pro
On-Demand

Creates production-ready frontend interfaces with bold aesthetics (glassmorphism, brutalism, neumorphism, etc.) and perfect images - real Unsplash/Pexels photos or custom AI image prompts.

Trigger Phrases:

  • "Create a stunning landing page"
  • "Design a modern interface"
  • "Build a professional UI"
  • Any frontend design request
11 aesthetic styles Unsplash/Pexels integration
💡 createsolution ↗ Docs
On-Demand - Workflow

Execute the AI Board Review workflow: refine requirements, analyze, gather multi-AI perspectives (Gemini, Codex), synthesize into a final solution, and create an executive web presentation.

Workflow Steps:

  • Refine program ask into formal spec
  • Submit to AI Review Board
  • Synthesize multi-model feedback
  • Create executive presentation site
$HOME/projects/ainotes/createsolution/ nginx deployment
📅 createplan ↗ Docs
On-Demand - Workflow

Multi-AI project planning workflow: all three AIs (Claude, Gemini, Codex) generate plans in parallel, synthesize into a draft, peer review, and create a final actionable project plan with executive presentation.

Workflow Steps:

  • Clarify and format plan ask
  • Parallel plan generation (3 AIs)
  • Synthesize into draft plan
  • Peer review from Gemini & Codex
  • Create final plan and presentation
$HOME/projects/ainotes/createplan/ nginx deployment
🧪 test-skill
On-Demand - Development

Test skill for verifying hot reload functionality. Modify while Claude Code is running to test changes appear without restart.

When to Use:

  • Testing skill hot reload
  • Verifying skill system functionality
Development only

Skill Anatomy

Directory Structure

~/.claude/skills/ ├── ai-communication/ │ └── SKILL.md ├── architecture-diagram-creator/ │ ├── SKILL.md │ ├── assets/templates/ │ └── references/ ├── claudemd-generator/ │ └── SKILL.md ├── createplan/ │ └── SKILL.md ├── createsolution/ │ └── SKILL.md ├── frontend-design-pro/ │ └── SKILL.md ├── gitlab-integration/ │ └── SKILL.md ├── naming-validator/ │ ├── SKILL.md │ ├── references/naming-standards.md │ └── scripts/validate_name.py ├── openmemory/ │ └── SKILL.md ├── secrets-security/ │ ├── SKILL.md │ ├── references/ │ └── scripts/ ├── skill-creator/ │ ├── SKILL.md │ └── scripts/ └── test-skill/ └── SKILL.md

Progressive Disclosure

  1. Metadata (name + description) - Always in context (~100 words per skill)
  2. SKILL.md body - Loaded when skill triggers (<5k words)
  3. Bundled resources - Loaded as needed by Claude (scripts, references, assets)

Result: 0 tokens at boot, only loaded when invoked

Comparison: User vs Project Skills

Aspect User-Level (~/.claude/) Project-Level (~/projects/.claude/)
Scope All projects, all sessions Infrastructure projects only
Count 12 skills 8 skills
Purpose General utilities, personal preferences, meta-tools Infrastructure deployment, service configuration
Examples createsolution, createplan, ai-communication, frontend-design-pro, naming-validator keycloak-setup, postgres-integration, traefik-setup

User-Level Skills Documentation

Part of the Claude Code Context Loading Architecture

Context Comparison | Claude CLI Configuration | LLM Memory | All Documentation

Created: 2025-11-26