codingbuddy-claude-plugin

CodingBuddy Claude Code Plugin

Version 5.5.0

Multi-AI Rules for consistent coding practices - PLAN/ACT/EVAL workflow, specialist agents, and reusable skills for systematic development.

Installation

# Via npm
npm install codingbuddy-claude-plugin

# Or via Claude Code
claude plugin add codingbuddy

Features

Workflow Modes

• PLAN: Design implementation approach with TDD
• ACT: Execute changes following quality standards
• EVAL: Evaluate code quality and suggest improvements
• AUTO: Autonomous PLAN → ACT → EVAL cycle

Commands (codingbuddy:* namespace)

All commands use the codingbuddy: namespace to avoid collisions with Claude Code built-ins.

• /codingbuddy:plan - Enter PLAN mode
• /codingbuddy:act - Enter ACT mode
• /codingbuddy:eval - Enter EVAL mode
• /codingbuddy:auto - Enter AUTO mode
• /codingbuddy:checklist - Generate contextual checklists
• /codingbuddy:buddy - Show project status and next actions

Tip: Type PLAN, ACT, EVAL, or AUTO as keywords for the fastest workflow entry.

Command Migration

Legacy bare commands (/plan, /act, /eval, /auto, /buddy, /checklist) are deprecated. They continue to work during the transition period, but all new commands use the codingbuddy:* namespace.

See Migration Guide for details.

Specialist Agents

35 AI agents for different domains:

• Security, Performance, Accessibility
• Architecture, Testing, Code Quality
• Frontend, Backend, DevOps
• And more...

Skills

Reusable workflows for consistent development:

• Test-Driven Development
• Systematic Debugging
• API Design
• Refactoring
• And more...

Feature Availability

The plugin works standalone with core features. MCP integration unlocks the full experience:

Feature	Standalone	With MCP
PLAN/ACT/EVAL/AUTO keyword triggers	✅	✅
Staged planning (Discover→Design→Plan)	✅	✅
Clarification gate (ambiguous prompts)	✅	✅
Permission forecast (prompt-aware)	✅	✅
Council scene (agent eye glyphs)	✅ ¹	✅
Specialist agent prompts	⚠️ ¹	✅
Slash commands (codingbuddy:*)	✅	✅
Dynamic checklists	—	✅
Context persistence across modes	—	✅
Rule search & impact reports	—	✅
Session briefings & recovery	—	✅

¹ Standalone features that read agent definitions require a local .ai-rules/ directory. Run npx codingbuddy-rules init to scaffold it. See #1216 for the standalone enhancement roadmap.

MCP Integration (Recommended)

Install the CodingBuddy MCP server to unlock the full feature set:

npm install -g codingbuddy

The MCP server provides:

• Dynamic checklists and specialist agent recommendations
• Context management for PLAN/ACT/EVAL workflow
• Rule search, impact reports, and session recovery

Architecture

packages/rules/.ai-rules/     ← Single source of truth (agents, skills, rules)
        ↓ (MCP protocol)
packages/claude-code-plugin/  ← Thin plugin (manifest + MCP configuration)

This architecture ensures:

• No duplication: All definitions live in one place
• DRY principle: Changes only need to be made once
• Single source of truth: packages/rules/.ai-rules/ is the canonical source

Documentation

• Core Rules
• Agents Guide
• Skills Guide

License

MIT