[BUG] Plan agent has incorrect description (duplicates Explore agent)

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?

## Description
The Plan agent description is identical to the Explore agent description in the Task tool's agent
type definitions.

## Current behavior
Both agents show:
Fast agent specialized for exploring codebases. Use this when you need to quickly find files by
patterns (eg. "src/components/**/*.tsx"), search code for keywords (eg. "API endpoints"), or answer
questions about the codebase (eg. "how do API endpoints work?"). When calling this agent, specify
the desired thoroughness level: "quick" for basic searches, "medium" for moderate exploration, or
"very thorough" for comprehensive analysis across multiple locations and naming conventions.

## Expected behavior
Plan agent should have its own description explaining its purpose and when to use it. Based on the
naming convention and agent organization, likely involves planning implementations or task
breakdowns.

## Impact
Users cannot distinguish between Plan and Explore agents. LLM cannot make informed decisions about
which agent to invoke since the descriptions are identical.

## Location
Task tool description, subagent_type parameter documentation

What Should Happen?

Plan agent should have a description that reflects its planning/design purpose, something like:

Specialized agent for planning implementation approaches and breaking down complex tasks into
executable steps. Use this when you need to design system architecture, evaluate implementation
alternatives, create task breakdowns, or plan feature development before coding. The agent analyzes
requirements, proposes approaches with tradeoffs, and structures work into concrete steps.

Or based on the ExitPlanMode tool reference to "plan mode":
Planning agent that breaks down implementation tasks into structured steps before execution. Use
when you need to design an implementation approach, evaluate architectural alternatives, or create a
detailed plan for complex features. Returns a structured plan with steps, dependencies, and
tradeoffs for user approval.

The key distinction from Explore:

• Explore: Discovers and understands existing codebase structure
• Plan: Designs and structures future implementation work

Error Messages/Logs

Steps to Reproduce

1. Start Claude Code CLI session

1. Ask Claude to list all available agent descriptions:

Show me the description of all agents in your context.

1. Observe the output for Plan and Explore agents

1. Compare the descriptions - they are identical

Actual output:
Explore

Fast agent specialized for exploring codebases. Use this when you need to quickly find files by
patterns (eg. "src/components/**/*.tsx"), search code for keywords (eg. "API endpoints"), or answer
questions about the codebase (eg. "how do API endpoints work?"). When calling this agent, specify
the desired thoroughness level: "quick" for basic searches, "medium" for moderate exploration, or
"very thorough" for comprehensive analysis across multiple locations and naming conventions.

Plan

Fast agent specialized for exploring codebases. Use this when you need to quickly find files by
patterns (eg. "src/components/**/*.tsx"), search code for keywords (eg. "API endpoints"), or answer
questions about the codebase (eg. "how do API endpoints work?"). When calling this agent, specify
the desired thoroughness level: "quick" for basic searches, "medium" for moderate exploration, or
"very thorough" for comprehensive analysis across multiple locations and naming conventions.

Alternative reproduction:

1. Examine the Task tool definition in Claude Code's system prompt configuration
2. Locate the subagent_type parameter's available agent types
3. Find the Plan agent description
4. Verify it duplicates Explore agent description

Source location: Task tool description, line containing Plan agent definition in the available
agent types list.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.0.30

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Xterm

Additional Information

_No response_