---

type: skill name: claude-code-management description: Unified management for Claude Code artifacts - extensions (agents, commands, skills, plugins, hooks) and configuration files (CLAUDE.md) using templates and a two-phase API. version: 0.4.0 tags:

• core
• management
• extensions
• configuration
• hooks

---

Claude Code Management

Unified management interface for Claude Code artifacts:

• Extensions: agents, commands, skills, plugins
• Hooks: lifecycle automation (settings.json configuration)
• Configuration: CLAUDE.md files

Activation

This skill activates when:

• User invokes /aida agent [create|validate|version|list]
• User invokes /aida command [create|validate|version|list]
• User invokes /aida skill [create|validate|version|list]
• User invokes /aida plugin [create|validate|version|list|add|remove]
• User invokes /aida hook [list|add|remove|validate]
• User invokes /aida claude [create|optimize|validate|list]
• Extension, hook, or configuration management is needed

Extension Operations

Command Routing

Parse the command to determine:

1. Component type: agent, command, skill, plugin, or hook
2. Operation: create, validate, version, list, add, remove
3. Arguments: name, description, options

Note: Hooks use different operations (see Hook Operations section).

Create Operations

For create operations, use the three-phase orchestration pattern:

Phase 1: Gather Context (Python)

Run the script to infer metadata and get questions:

Copypython {base_directory}/scripts/manage.py --get-questions \
  --context='{"operation": "create", "type": "agent", "description": "user description", "location": "user"}'

Returns:

Copy{
  "inferred": {
    "name": "inferred-name",
    "version": "0.1.0",
    "tags": ["custom"],
    "base_path": "~/.claude/agents/inferred-name"
  },
  "questions": [
    {"id": "location", "question": "Where to create?", "options": [...]}
  ],
  "project_context": {
    "languages": ["python"],
    "frameworks": ["fastapi"]
  }
}

Phase 2: Generate Content (Agent)

Spawn the claude-code-expert agent with complete context and output contract:

CopyOperation: CREATE
Type: agent

Requirements:
  - Name: {inferred.name}
  - Description: {user_description}
  - Location: {inferred.base_path}
  - Project Context: {project_context}
  - User Answers: {answers to questions}

Output Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": true|false,
    "issues": [
      {"severity": "error|warning", "message": "description", "suggestion": "how to fix"}
    ]
  },
  "files": [
    {
      "path": "relative/path/to/file.md",
      "content": "complete file content as string"
    }
  ],
  "summary": {
    "created": ["list of files"],
    "next_steps": ["what user should do next"]
  }
}

If validation fails (passed=false with errors), return empty files array.
Warnings should not prevent file generation.

Handle agent response:

1. Parse the JSON response
2. If validation.passed is false with errors:
   • Report issues to user
   • Ask if they want to proceed anyway or provide more info
3. If validation passed (or user chose to proceed):
   • Continue to Phase 3

Phase 3: Write Files (Python)

Pass agent output to Python for file creation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "create", "type": "agent", "agent_output": <agent_json>}'

The script:

1. Validates file structure (required fields, proper frontmatter)
2. Creates directories
3. Writes files
4. Returns success/failure

Validate Operations

For validate operations:

1. Run validation script
2. Report results with any errors

Script invocation:

Copy# Validate specific component
python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "validate", "type": "agent", "name": "my-agent"}'

# Validate all
python {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "validate", "type": "agent", "all": true, "location": "all"}'

Version Operations

For version operations:

1. Find the component
2. Bump version (major, minor, or patch)
3. Update the file
4. Report the version change

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "version", "type": "agent", "name": "my-agent", "bump": "patch"}'

List Operations

For list operations:

1. Search specified locations
2. Parse component metadata
3. Return formatted list

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"operation": "list", "type": "agent", "location": "all", "format": "table"}'

Plugin-Specific Operations

For add and remove on plugins:

• add: Create component inside plugin directory
• remove: Remove component from plugin (with confirmation)

Agent Output Contract

When spawning claude-code-expert for CREATE operations, always include this output format specification:

For Agents

CopyOutput Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": boolean,
    "issues": [
      {"severity": "error|warning", "message": "string", "suggestion": "string"}
    ]
  },
  "files": [
    {"path": "agents/{name}/{name}.md", "content": "..."},
    {"path": "agents/{name}/knowledge/index.md", "content": "..."},
    {"path": "agents/{name}/knowledge/{topic}.md", "content": "..."}
  ],
  "summary": {
    "created": ["list of relative paths"],
    "next_steps": ["actionable items for user"]
  }
}

For Commands

CopyOutput Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": boolean,
    "issues": [...]
  },
  "files": [
    {"path": "commands/{name}.md", "content": "..."}
  ],
  "summary": {
    "created": ["list of relative paths"],
    "next_steps": ["actionable items for user"]
  }
}

For Skills

CopyOutput Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": boolean,
    "issues": [...]
  },
  "files": [
    {"path": "skills/{name}/SKILL.md", "content": "..."},
    {"path": "skills/{name}/scripts/{script}.py", "content": "..."},
    {"path": "skills/{name}/references/{doc}.md", "content": "..."},
    {"path": "skills/{name}/templates/{template}.jinja2", "content": "..."}
  ],
  "summary": {
    "created": ["list of relative paths"],
    "next_steps": ["actionable items for user"]
  }
}

For Plugins

CopyOutput Format:
Return a JSON object with this exact structure:

{
  "validation": {
    "passed": boolean,
    "issues": [...]
  },
  "files": [
    {"path": ".claude-plugin/plugin.json", "content": "..."},
    {"path": "README.md", "content": "..."},
    {"path": ".gitignore", "content": "..."},
    {"path": "agents/{name}/{name}.md", "content": "..."}
  ],
  "summary": {
    "created": ["list of relative paths"],
    "next_steps": ["actionable items for user"]
  }
}

Hook Operations

Hooks are different from other extensions - they're JSON configuration in settings.json, not separate files. Operations are simpler and more focused.

Understanding Hooks

Hooks are lifecycle automation that execute shell commands at specific events:

• PreToolUse / PostToolUse - Before/after tool execution
• SessionStart / SessionEnd - Session lifecycle
• UserPromptSubmit - When user submits prompt
• Notification / Stop - Response lifecycle

Hooks live in settings.json files:

• User: ~/.claude/settings.json
• Project: ./.claude/settings.json
• Project local: ./.claude/settings.local.json

List Operations

For list operations:

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "hook", "operation": "list", "scope": "all"}'

Scopes:

• user - User-level hooks only
• project - Project-level hooks only
• all - All hooks (default)

Output: Table showing hooks with event, matcher, command, and source file.

Add Operations

For add operations, use a two-phase pattern:

Phase 1: Gather Context

Copypython {base_directory}/scripts/manage.py --get-questions \
  --context='{"target": "hook", "operation": "add", "description": "user description"}'

Returns questions about:

• Event type (PreToolUse, PostToolUse, etc.)
• Matcher pattern (tool names)
• Scope (user vs project)
• Whether to use a template (formatter, logger, blocker)

Phase 2: Execute

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "hook", "operation": "add", "event": "PostToolUse", "matcher": "Write|Edit", "command": "prettier --write", "scope": "project"}'

Common templates available:

Remove Operations

For remove operations:

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "hook", "operation": "remove", "event": "PostToolUse", "matcher": "Write|Edit", "scope": "project"}'

Removes the matching hook from the settings file.

Validate Operations

For validate operations:

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "hook", "operation": "validate", "scope": "all"}'

Validates:

• Hook structure (required fields)
• Event names are valid
• Commands are safe (warns about dangerous patterns)
• No duplicate hooks

CLAUDE.md Operations

Command Routing

Parse the command to determine:

1. Operation: create, optimize, validate, list
2. Scope: project, user, plugin, or all
3. Arguments: path, options

Create Operations

For create operations:

1. Read references/claude-md-workflow.md for the full workflow
2. Run Phase 1 to detect project context
3. Ask user scope if not specified
4. Run Phase 2 to create the file
5. Report success

Script invocation:

Copy# Phase 1: Get questions
python {base_directory}/scripts/manage.py --get-questions \
  --context='{"target": "claude", "operation": "create", "scope": "project"}'

# Phase 2: Execute
python {base_directory}/scripts/manage.py --execute \
  --context='{"target": "claude", "operation": "create", "scope": "project", "name": "...", "description": "...", "commands": [...]}'

Optimize Operations

For optimize operations:

1. Read references/best-practices.md for scoring criteria
2. Run Phase 1 to audit and generate findings
3. Ask user how to fix (all, critical only, interactive, skip)
4. Run Phase 2 to apply fixes
5. Report new score and changes

Script invocation:

Copy# Phase 1: Get audit and questions
python {base_directory}/scripts/manage.py --get-questions \
  --context='{"target": "claude", "operation": "optimize", "scope": "project"}'

# Phase 2: Apply fixes
python {base_directory}/scripts/manage.py --execute \
  --context='{"target": "claude", "operation": "optimize", "scope": "project", "fix_mode": "Fix all"}'

Validate Operations

For validate operations:

1. Find CLAUDE.md files in specified scope
2. Run validation checks
3. Report results

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "claude", "operation": "validate", "scope": "all"}'

List Operations

For list operations:

1. Find all CLAUDE.md files
2. Include validation status
3. Return formatted list

Script invocation:

Copypython {base_directory}/scripts/manage.py --execute \
  --context='{"target": "claude", "operation": "list", "scope": "all"}'

Path Resolution

Base Directory: Provided when skill loads via <command-message> tags.

Script Execution:

Copy{base_directory}/scripts/manage.py

Reference Files:

Copy{base_directory}/references/create-workflow.md
{base_directory}/references/validate-workflow.md
{base_directory}/references/schemas.md
{base_directory}/references/claude-md-workflow.md
{base_directory}/references/best-practices.md

Location Options

Extension Locations

CLAUDE.md Scopes

Hook Scopes

Example Workflows

Creating an Agent (Full Flow)

CopyUser: /aida agent create "handles database migrations"

1. Parse: type=agent, operation=create, description="handles database migrations"

2. Phase 1 (Python):
   python manage.py --get-questions --context='{...}'
   Returns:
   - inferred: name="database-migration", version="0.1.0"
   - questions: [location question]
   - project_context: {languages: ["python"], frameworks: ["alembic"]}

3. Ask user questions (if any):
   AskUserQuestion: "Where should we create this agent?"

4. Phase 2 (Agent):
   Spawn claude-code-expert with:
   - Operation: CREATE
   - Type: agent
   - Name: database-migration
   - Description: handles database migrations
   - Location: ~/.claude/agents/database-migration
   - Project Context: Python, Alembic detected
   - Output Format: [JSON contract]

   Agent returns:
   {
     "validation": {"passed": true, "issues": []},
     "files": [
       {"path": "agents/database-migration/database-migration.md", "content": "..."},
       {"path": "agents/database-migration/knowledge/index.md", "content": "..."},
       {"path": "agents/database-migration/knowledge/migration-patterns.md", "content": "..."}
     ],
     "summary": {
       "created": ["database-migration.md", "knowledge/index.md", "knowledge/migration-patterns.md"],
       "next_steps": ["Review migration-patterns.md and add project-specific patterns"]
     }
   }

5. Phase 3 (Python):
   python manage.py --execute --context='{"operation": "create", "agent_output": {...}}'
   - Validates structure
   - Creates directories
   - Writes files

6. Report to user:
   "Created agent 'database-migration' with 3 files:
    - database-migration.md
    - knowledge/index.md
    - knowledge/migration-patterns.md

   Next steps:
    - Review migration-patterns.md and add project-specific patterns"

Optimizing CLAUDE.md

CopyUser: /aida claude optimize

1. Parse: target=claude, operation=optimize, scope=project
2. Run Phase 1:
   - Find CLAUDE.md
   - Audit against best practices
   - Score: 65/100
   - Findings: 3 issues (1 critical)
3. Ask: "How would you like to fix them?"
4. User: "Fix all"
5. Run Phase 2:
   - Apply fixes
   - New score: 85/100
6. Report: "Applied 3 fixes. Score improved from 65 to 85."

Resources

scripts/

• manage.py - Main dispatcher script
• operations/ - Operation modules
  • utils.py - Shared utilities
  • extensions.py - Extension operations (agent/command/skill/plugin)
  • claude_md.py - CLAUDE.md operations
  • hooks.py - Hook operations (list/add/remove/validate)

references/

• create-workflow.md - Extension create workflow (deprecated - use this SKILL.md)
• validate-workflow.md - Extension validation rules
• schemas.md - Frontmatter schema reference
• claude-md-workflow.md - CLAUDE.md create/optimize workflow
• best-practices.md - CLAUDE.md best practices and scoring