Ascentic.SkillKit 1.0.4

Ascentic.SkillKit

A generic dotnet tool that installs AI skills and agents into your projects. Works with any AI assistant (Claude, ChatGPT, GitHub Copilot, etc.) by creating a customizable folder structure.

Install from NuGet

dotnet tool install --global Ascentic.SkillKit

Quick Start

# Install all skills & agents into .claude directory
skill-kit init .claude

# Install only Clean Architecture skills into .ai directory
skill-kit init .ai --profile cleanarch

# List all available skills and agents
skill-kit list

Commands

Core Commands

init - Initialize skills & agents folder

Install skills and agents into your project with a custom directory name.

# Install all skills & agents into .claude directory
skill-kit init .claude

# Install specific profile into .ai directory
skill-kit init .ai --profile cleanarch
skill-kit init .skills --profile vertslice
skill-kit init .copilot --profile modmonolith

# Target specific directory
skill-kit init .claude --output /path/to/project

# Force overwrite existing folder
skill-kit init .claude --force

Arguments:

• directory - Directory name for skills & agents (e.g., .claude, .ai, .skills) - REQUIRED

Options:

• --output, -o - Target directory (defaults to current directory)
• --profile, -p - Architecture profile: all, cleanarch, vertslice, modmonolith
• --force, -f - Overwrite existing folder

create-doc - Create claude.md documentation file

Generate a claude.md file for your project.

# Create claude.md at the same level as .claude directory (sibling - default)
skill-kit create-doc .claude

# Create claude.md inside the .claude directory
skill-kit create-doc .claude --location inside

# Create in specific directory
skill-kit create-doc .ai --output /path/to/project

# Force overwrite existing file
skill-kit create-doc .claude --force

# Combine options
skill-kit create-doc .claude -l inside -o /path/to/project -f

Arguments:

• directory - Directory name containing skills & agents (e.g., .claude, .ai) - REQUIRED

Options:

• --location, -l - Where to place claude.md: sibling (same level as directory, default) or inside (within directory)
• --output, -o - Target directory (defaults to current directory)
• --force, -f - Overwrite existing claude.md if present

Location Options Explained:

Location	File Path	Use Case
sibling (default)	project/claude.md	Documentation visible at project root, easy to discover
inside	project/.claude/claude.md	Self-contained setup, all AI files in one directory

Sibling Location (Default):

your-project/
├── .claude/
│   ├── skills/
│   └── agents/
├── claude.md          ← Created here
└── README.md

Inside Location:

your-project/
├── .claude/
│   ├── skills/
│   ├── agents/
│   └── claude.md      ← Created here
└── README.md

Generated Content Includes:

• Project overview and structure
• Skills documentation links
• Agents documentation links
• Usage guidelines and best practices
• Resource references

When to Use Each Location:

Use sibling (default) when:

• You want documentation easily discoverable at project root
• Following common project conventions
• Working on open source projects
• Team expects docs at top level

Use inside when:

• You prefer a self-contained AI setup
• Project root is already crowded
• Packaging the directory for distribution
• Want to keep AI documentation private/internal

list - List available skills and agents

Display all skills and agents grouped by architecture profile.

skill-kit list

update - Update existing folder

Update your folder with latest skills and agents.

# Merge mode: add new items only (default)
skill-kit update .claude

# Overwrite mode: replace all matching items
skill-kit update .ai --overwrite

# Update specific profile
skill-kit update .claude --profile cleanarch

# Update in specific directory
skill-kit update .claude --output /path/to/project

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Options:

• --output, -o - Target directory
• --merge, -m - Keep existing files, only add new ones (default)
• --overwrite - Replace all matching skills/agents with latest versions
• --profile, -p - Architecture profile to update

MCP Server Commands

Transform your static skills folder into an active, intelligent "Memory Bank" and architectural orchestrator by connecting skill-kit directly to AI tools (like Claude Code) via the Model Context Protocol (MCP).

mcp setup - Configure AI Integration

Generate a project-level .mcp.json file that automatically connects your skill-kit skills and project memory to Claude Code.

# Setup MCP connection for Claude Code referencing the .claude directory
skill-kit mcp setup .claude

# Target a specific external directory
skill-kit mcp setup .claude --output /path/to/project

# Overwrite existing .mcp.json file
skill-kit mcp setup .claude --force

What this enables for your AI:

• Dynamic Memory: Claude can actively write (memory_store) and read (memory_search) architectural decisions specific to your project, so it automatically remembers your rules tomorrow.
• Skill Awareness: Claude natively lists installed skills (skill_list) and fetches your strict scaffold instructions instantly (knowledge_read) instead of hallucinating.

mcp start - Start the MCP Server

Typically executed automatically by your AI assistant using the generated .mcp.json file. It spins up the skill-kit JSON-RPC server over standard input/output.

# Start the MCP server explicitly (blocks terminal)
skill-kit mcp start .claude

Management Commands

add - Add specific skills or agents

Add individual skills or agents to existing folder.

# Add single skill
skill-kit add .claude --skill dotnet-cleanarch-ddd-api-create-endpoint

# Add multiple skills
skill-kit add .ai --skill skill1 --skill skill2

# Add agent
skill-kit add .claude --agent code-reviewer

# Add to specific directory
skill-kit add .claude --skill skill-name --output /path/to/project

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Options:

• --skill, -s - Skill name(s) to add
• --agent, -a - Agent name(s) to add
• --output, -o - Target directory

remove - Remove specific skills or agents

Remove skills or agents from folder.

# Remove skill
skill-kit remove .claude --skill dotnet-cleanarch-ddd-api-create-endpoint

# Remove multiple items
skill-kit remove .ai --skill skill1 --skill skill2 --agent agent1

# Remove from specific directory
skill-kit remove .claude --skill skill-name --output /path/to/project

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Options:

• --skill, -s - Skill name(s) to remove
• --agent, -a - Agent name(s) to remove
• --output, -o - Target directory

Information Commands

info - Show detailed information

Display detailed information about a specific skill or agent.

# Get skill info
skill-kit info --skill dotnet-cleanarch-ddd-api-create-endpoint

# Get agent info
skill-kit info --agent cleanarch-ddd-guardian

Options:

• --skill, -s - Skill name to get info about
• --agent, -a - Agent name to get info about

search - Search for skills and agents

Search for skills and agents by keyword.

# Search by keyword
skill-kit search "repository"
skill-kit search "api endpoint"
skill-kit search "validation"

diff - Compare local with registry

Compare your local folder with the registry to see differences.

# Show diff for .claude directory
skill-kit diff .claude

# Show diff for specific directory
skill-kit diff .ai --output /path/to/project

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Shows:

• Installed skills/agents from registry
• Missing items (in registry but not local)
• Custom items (local but not in registry)

Options:

• --output, -o - Target directory

Validation & Diagnostics

validate - Validate folder

Check folder structure and validate all skills and agents.

# Validate .claude directory
skill-kit validate .claude

# Validate specific directory
skill-kit validate .ai --output /path/to/project

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Checks:

• Missing required files (skill.md, agent.md)
• Empty files
• Folder structure

Options:

• --output, -o - Target directory

doctor - Run health check

Run diagnostics to check tool installation health.

skill-kit doctor

Checks:

• Tool version
• .NET runtime version
• Embedded resources integrity
• Skill registry
• File system permissions

Import/Export Commands

export - Export folder

Export your folder as a shareable zip package.

# Export from .claude directory
skill-kit export .claude

# Export with custom output name
skill-kit export .ai --output my-skills.zip

# Export from specific directory
skill-kit export .claude --source /path/to/project --output skills.zip

Arguments:

• directory - Directory name containing skills & agents - REQUIRED

Options:

• --source, -s - Source directory containing the folder
• --output, -o - Output zip file path (defaults to skills-export.zip)

import - Import skills and agents

Import skills and agents from a zip file or URL.

# Import from local zip file to .claude
skill-kit import .claude --source custom-skills.zip

# Import from URL to .ai
skill-kit import .ai --url https://example.com/skills.zip

# Import to specific directory
skill-kit import .claude --source skills.zip --output /path/to/project

Arguments:

• directory - Directory name for skills & agents - REQUIRED

Options:

• --source, -s - Path to zip file
• --url, -u - URL to download zip file from
• --output, -o - Target directory

Profiles

Profile	Includes
cleanarch	Clean Architecture / DDD skills + guardians
vertslice	Vertical Slice skills + vertslice-guardian
modmonolith	Modular Monolith skills
all	Everything (default)

Output Structure

your-project/
└── .claude/  (or .ai, .skills, etc.)
    ├── skills/
    │   ├── dotnet-cleanarch-ddd-api-create-endpoint/
    │   │   └── skill.md
    │   ├── dotnet-vertslice-feature-create-slice/
    │   │   └── skill.md
    │   └── ... (26 skills total)
    └── agents/
        ├── cleanarch-ddd-guardian/
        │   └── agent.md
        ├── code-reviewer/
        │   └── agent.md
        └── ... (5 agents total)

Common Workflows

Initial Setup

# Install tool
dotnet tool install --global Ascentic.SkillKit

# Initialize project with Clean Architecture skills
skill-kit init .claude --profile cleanarch

# Create documentation file at project root
skill-kit create-doc .claude

Creating Documentation

# Standard setup - documentation at project root (recommended for most projects)
skill-kit create-doc .claude

# Self-contained setup - documentation inside .claude directory
skill-kit create-doc .claude --location inside

# Update existing documentation
skill-kit create-doc .claude --force

# Create for different AI directory
skill-kit create-doc .ai --location inside

# Create for project in different location
skill-kit create-doc .claude --output /path/to/project

Complete Workflow with Documentation:

# 1. Initialize with specific profile
skill-kit init .claude --profile cleanarch

# 2. Create documentation
skill-kit create-doc .claude

# 3. Validate setup
skill-kit validate .claude

# 4. View what's installed
skill-kit list

Adding More Skills Later

# Add a specific skill
claude-kit add --skill dotnet-cleanarch-ddd-domain-create-aggregate

# Add multiple skills
claude-kit add --skill skill1 --skill skill2 --agent code-reviewer

Keeping Skills Updated

# Check what's different
claude-kit diff

# Update to latest (merge mode - keeps custom skills)
claude-kit update

# Force update everything
claude-kit update --overwrite

Sharing Custom Skills

# Export your customized .claude folder
claude-kit export --output my-team-skills.zip

# Team member imports it
claude-kit import --source my-team-skills.zip

Managing Documentation

# Create initial documentation
skill-kit create-doc .claude

# Update documentation after adding new skills
skill-kit add .claude --skill new-skill-name
skill-kit create-doc .claude --force

# Switch documentation location
skill-kit create-doc .claude --location inside --force

# Create documentation for multiple AI directories
skill-kit create-doc .claude
skill-kit create-doc .copilot
skill-kit create-doc .ai

Multi-Project Setup

# Setup for monorepo with multiple services
for service in api web worker; do
  skill-kit init .claude --output ./services/$service --profile cleanarch
  skill-kit create-doc .claude --output ./services/$service --location inside
done

# Setup for different AI tools in same project
skill-kit init .claude --profile cleanarch
skill-kit create-doc .claude

skill-kit init .copilot --profile vertslice
skill-kit create-doc .copilot

Troubleshooting

# Check tool health
claude-kit doctor

# Validate your .claude folder
claude-kit validate

# Search for a specific skill
claude-kit search "repository"

# Get detailed info about a skill
claude-kit info --skill dotnet-cleanarch-ddd-api-create-endpoint

Real-World Examples

Example 1: New Clean Architecture Project

# Setup
dotnet tool install --global Ascentic.SkillKit
cd my-new-project

# Initialize with Clean Architecture skills
skill-kit init .claude --profile cleanarch

# Create documentation at project root
skill-kit create-doc .claude

# Validate everything is correct
skill-kit validate .claude

# Result: Ready to use AI skills with documentation

Example 2: Existing Project - Add AI Capabilities

# Navigate to existing project
cd existing-project

# Add all skills and agents
skill-kit init .ai --profile all

# Create documentation inside .ai directory (keep root clean)
skill-kit create-doc .ai --location inside

# Check what was installed
skill-kit list

Example 3: Team Collaboration

# Team lead: Setup and export
skill-kit init .claude --profile cleanarch
skill-kit add .claude --skill custom-skill-1 --skill custom-skill-2
skill-kit create-doc .claude
skill-kit export .claude --output team-ai-setup.zip

# Team member: Import and use
skill-kit import .claude --source team-ai-setup.zip
# Documentation is already included in the export!

Example 4: Multiple Architecture Styles

# Setup different profiles for comparison
skill-kit init .cleanarch --profile cleanarch
skill-kit create-doc .cleanarch

skill-kit init .vertslice --profile vertslice
skill-kit create-doc .vertslice

skill-kit init .modmonolith --profile modmonolith
skill-kit create-doc .modmonolith

# Compare and choose the best fit for your project

Example 5: Microservices Architecture

# Setup each microservice with its own AI skills
services=("auth" "users" "payments" "notifications")

for service in "${services[@]}"; do
  echo "Setting up $service..."
  skill-kit init .claude --output ./services/$service --profile cleanarch
  skill-kit create-doc .claude --output ./services/$service --location inside
done

# Each service now has self-contained AI documentation

Example 6: Continuous Updates

# Initial setup
skill-kit init .claude --profile cleanarch
skill-kit create-doc .claude

# Later: Check for updates
skill-kit diff .claude

# Update skills (keeps custom additions)
skill-kit update .claude

# Refresh documentation
skill-kit create-doc .claude --force

# Validate everything still works
skill-kit validate .claude

Update / Uninstall

# Update to latest version
dotnet tool update --global Ascentic.SkillKit

# Uninstall
dotnet tool uninstall --global Ascentic.SkillKit

---

Development Guide (Maintaining This Tool)

Prerequisites

• .NET 9 SDK
• PowerShell (Windows) or Bash (Mac/Linux)

Project Structure

Ascentic.SkillKit/
├── .claude/                          ← Your actual skills & agents live here
│   ├── skills/
│   │   ├── dotnet-cleanarch-ddd-api-create-endpoint/
│   │   │   └── skill.md  (+ any subdirs)
│   │   └── ...
│   └── agents/
│       ├── cleanarch-ddd-guardian/
│       │   └── agent.md
│       └── ...
├── src/Ascentic.SkillKit/
│   ├── Ascentic.SkillKit.csproj
│   ├── Program.cs
│   ├── Commands/
│   │   ├── InitCommand.cs
│   │   ├── ListCommand.cs
│   │   ├── UpdateCommand.cs
│   │   └── ... (all command files)
│   ├── Services/
│   │   ├── ScaffolderService.cs
│   │   ├── SkillRegistry.cs
│   │   └── ... (all service files)
│   └── Resources/
│       └── claude.zip                ← Generated by pack-resources script
├── scripts/
│   ├── pack-resources.ps1            ← Windows
│   └── pack-resources.sh             ← Mac/Linux
└── Ascentic.SkillKit.sln

Step 1 — Edit Your Skills/Agents

All content lives in .claude/ at the repo root. Edit files there freely. If you add or remove a skill/agent folder, also update SkillRegistry.cs.

Step 2 — Pack Resources

# Windows
pwsh scripts\pack-resources.ps1

# Mac/Linux
./scripts/pack-resources.sh

This zips .claude/ → src/Ascentic.SkillKit/Resources/claude.zip

Step 3 — Build & Test Locally

dotnet build
dotnet pack src/Ascentic.SkillKit --configuration Release --output ./nupkg
dotnet tool uninstall --global Ascentic.SkillKit
dotnet tool install --global --add-source ./nupkg Ascentic.SkillKit

# Test commands
skill-kit init .claude --output /tmp/test-project
skill-kit list
skill-kit doctor

Step 4 — Publish to NuGet

Push a version tag — GitHub Actions handles the rest:

git tag v1.0.1
git push origin v1.0.1

Add NUGET_API_KEY to your GitHub repo secrets first.

License

MIT