[BUG] Plugin Redundancies/Conflicts

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

The included commands and agents in the your plugin repo are conflicting/overlapping/redundant, potentially because of an oversight of the poor coordination between code owners.

What Should Happen?

A build process should detect this automatically and flag the problems.

Error Messages/Logs

It was quite a lot but the code review came out strongly

Code Review (4 overlapping mechanisms)
  - Skill: code-review
  - Command: /code-review:code-review
  - Agent: pr-review-toolkit:code-reviewer
  - Agent: feature-dev:code-reviewer

  I already read the first three. Let me compare them:

  code-review skill:
  - Educational/protocol guide
  - Three practices: receiving feedback, requesting reviews, verification gates
  - References to use code-reviewer subagent
  - Not executable itself - teaches when/how to use review

  /code-review:code-review command:
  - Automated PR review workflow
  - Launches 5 parallel Sonnet agents
  - Checks CLAUDE.md compliance, bugs, history, previous PRs, comments
  - Confidence scoring (0-100), filters <80
  - Posts review comment via gh

  pr-review-toolkit:code-reviewer agent:
  - Opus model
  - Reviews code against CLAUDE.md
  - Confidence scoring ≥80
  - For proactive review after writing code

  feature-dev:code-reviewer agent:
  - Sonnet model
  - Same purpose as pr-review-toolkit version
  - Almost identical system prompt

...

Code Review Analysis

  1. code-review skill (user space):
  - Type: Educational guide/protocol
  - Purpose: Teaches WHEN and HOW to use code review
  - Not executable - references other tools

  2. /code-review:code-review command (plugin):
  - Type: Automated PR review workflow
  - Purpose: Complex multi-agent PR review on GitHub
  - Scope: Pull requests specifically
  - Uses: 5 parallel agents, CLAUDE.md checks, gh CLI

  3. pr-review-toolkit:code-reviewer agent:
  - Type: Opus agent for proactive review
  - Purpose: Review code after writing
  - Scope: Local changes (unstaged git diff)
  - Model: Opus
  - Tools: None (relies on context)

  4. feature-dev:code-reviewer agent:
  - Type: Sonnet agent for proactive review
  - Purpose: Identical to #3
  - Scope: Local changes (unstaged git diff)
  - Model: Sonnet
  - Tools: Full toolkit (Glob, Grep, Read, etc.)

  ---
  The Redundancy

  #3 and #4 are 95% identical:
  - Same system prompt (word-for-word)
  - Same purpose (review code after writing)
  - Same scope (unstaged changes)
  - Same confidence threshold (≥80)
  - Only differences:
    - Model: Opus vs Sonnet
    - Tools: None vs Full toolkit
    - Color: Green vs Red
    - Description verbosity

  These are duplicate agents in different plugins.

  ---
  Recommendation

  Delete feature-dev:code-reviewer, keep pr-review-toolkit:code-reviewer

  Why keep pr-review-toolkit version:
  - ✅ Part of focused PR review toolkit
  - ✅ Uses Opus (higher quality for review)
  - ✅ More detailed description with examples
  - ✅ Clearer triggering conditions

  Why delete feature-dev version:
  - ❌ Exact duplicate functionality
  - ❌ Uses Sonnet (faster but lower quality)
  - ❌ Has tools it doesn't need (the prompt doesn't use them)
  - ❌ Unclear why feature-dev plugin needs its own reviewer

  Non-redundant items:
  - ✅ Keep code-review skill (teaches protocol)
  - ✅ Keep /code-review:code-review command (automated PR workflow)
  - ✅ Keep pr-review-toolkit:code-reviewer agent (proactive review)

  Should I delete feature-dev:code-reviewer?

Steps to Reproduce

Ask Claude Code the following question after enabling all the marketplace plugins in a new project:

which plugins, skills or commands that you just been given are conflicting or redundant?. ultrathink

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.0.50

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

_No response_