Table of Contents

Skills, Plugins, and Hooks

JD.AI integrates with the Claude Code extension ecosystem. Skills, plugins, and hooks are loaded at startup and registered as Semantic Kernel functions and filters through the JD.SemanticKernel.Extensions library.

Skills

Skills are markdown files (SKILL.md) that provide instructions and context to the AI agent. Each skill declares a name, description, and optional tool constraints in YAML frontmatter.

Where skills live

Location Path Scope
Personal ~/.claude/skills/<name>/SKILL.md All projects
Project .claude/skills/<name>/SKILL.md This project only

Project skills take precedence when names collide with personal skills.

How JD.AI loads skills

At startup, JD.AI:

  1. Scans ~/.claude/skills/ for personal skills
  2. Scans .claude/skills/ in the project directory
  3. Parses SKILL.md frontmatter (name, description, allowed-tools)
  4. Registers each skill as available context for the agent

Example skill

~/.claude/skills/code-review/SKILL.md

---
name: code-review
description: Review code for quality and best practices
---

When reviewing code:
1. Check for error handling
2. Verify input validation
3. Look for security vulnerabilities
4. Ensure test coverage

Startup output

When skills load successfully, JD.AI logs each one:

Loaded skills from C:\Users\you\.claude\skills
  • code-review: Review code for quality and best practices
  • commit-message: Generate conventional commit messages

Plugins

Plugins are directories with a plugin.json manifest that bundle skills, hooks, and configuration into a single distributable unit.

Plugin structure

.claude/plugins/my-plugin/
├── plugin.json              # Manifest
├── skills/
│   └── my-skill/SKILL.md   # Plugin skills
└── hooks/
    └── hooks.json           # Plugin hooks

Loading

Location Path
Personal ~/.claude/plugins/<name>/
Project .claude/plugins/<name>/

JD.AI reads each plugin.json, resolves internal dependencies, and registers the plugin's skills and hooks in a single pass.

Hooks

Hooks are event-driven filters that run before or after tool execution.

Hook events

Event When
PreToolUse Before a tool is invoked
PostToolUse After a tool completes

How hooks integrate

Hooks are parsed from hooks.json and registered as Semantic Kernel IFunctionInvocationFilter instances. They can:

  • Modify arguments before execution
  • Block execution based on rules
  • Post-process results after a tool completes
  • Log or audit tool usage

How it connects to Semantic Kernel

JD.AI uses JD.SemanticKernel.Extensions to bridge Claude Code's file-based extension format into Semantic Kernel primitives:

Claude Code concept Semantic Kernel equivalent
SKILL.md KernelFunction
hooks.json IFunctionInvocationFilter / IPromptRenderFilter
plugin.json Plugin with dependency resolution

This mapping lets the same skill and hook files work in both Claude Code and JD.AI without modification.

Package Description
JD.SemanticKernel.Extensions Core parsing and registration of skills, plugins, and hooks
JD.SemanticKernel.Connectors.ClaudeCode Claude Code authentication provider
JD.SemanticKernel.Connectors.GitHubCopilot GitHub Copilot authentication provider
JD.SemanticKernel.Connectors.OpenAICodex OpenAI Codex authentication provider
Edit this page