Claude Code Plugins

Community-maintained marketplace

Feedback

Use this skill when working on code projects that need memory persistence, code validation, multi-agent coordination, or workflow tracking. Activates Claude Code Guardian MCP tools.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name ccg-guardian
description Use this skill when working on code projects that need memory persistence, code validation, multi-agent coordination, or workflow tracking. Activates Claude Code Guardian MCP tools.
allowed-tools mcp__code-guardian__*

CCG Guardian Skill

Enterprise-grade development assistance with persistent memory, code validation, and multi-agent coordination.

When to Use

Activate this skill when:

  • Starting a new development session (call session_init)
  • Need to remember decisions, patterns, or context across sessions
  • Want code validation before commits (security, quality checks)
  • Working on multi-file tasks that need progress tracking
  • Need specialized agent guidance (Trading, Laravel, React, Node.js)

Core Capabilities

1. Memory System

memory_store     - Save decisions, facts, patterns (importance 1-10)
memory_recall    - Search memories by query, tags, type
memory_list      - List all memories with filters

2. Guard System (10 Security Rules)

guard_validate   - Check code for vulnerabilities
guard_check_test - Detect fake tests without assertions
guard_rules      - List all validation rules

Rules include: SQL Injection, XSS, Command Injection, Path Traversal, Prompt Injection, Hardcoded Secrets, Empty Catch, Disabled Features, Fake Tests, Emoji in Code

3. Workflow & Tasks

workflow_task_create   - Create task with priority/tags
workflow_task_start    - Begin working on task
workflow_task_complete - Mark task done
workflow_current       - Get current active task

4. Multi-Agent System

agents_list      - List specialized agents
agents_select    - Auto-select best agent for task
agents_coordinate - Coordinate multiple agents

Available Agents:

  • Trading Agent - Quant systems, risk management, backtesting
  • Laravel Agent - PHP backend, Eloquent, REST APIs
  • React Agent - Components, hooks, state management
  • Node Agent - Event-driven, workers, orchestration

5. Resource Management

resource_status           - Token usage and checkpoints
resource_checkpoint_create - Save current progress
resource_checkpoint_restore - Restore from checkpoint

Workflow Pattern

1. session_init          → Load memory, check status
2. workflow_task_create  → Define what you're doing
3. memory_recall         → Get relevant context
4. [do work]
5. guard_validate        → Check code quality
6. memory_store          → Save important decisions
7. workflow_task_complete → Mark done
8. session_end           → Save all data

Best Practices

  1. Always start with session_init - Loads previous context
  2. Store important decisions - Use importance 8-10 for critical items
  3. Validate before commit - Run guard_validate on changed files
  4. Track progress - Use workflow tasks for multi-step work
  5. End sessions properly - Call session_end to persist data

Example Usage

"Use CCG to help me refactor the authentication module with proper memory and validation"

Claude will:

  1. Initialize session and recall relevant memories
  2. Create a task for tracking
  3. Select appropriate agent (Node or Laravel)
  4. Apply changes with guard validation
  5. Store decisions for future reference