Skip to main content
Recreates the deprecated Explanatory output style as a SessionStart hook, providing educational context about your code.
This plugin adds additional instructions and output to every session, which will increase token usage and costs.

​
Overview

The Explanatory Output Style plugin automatically enhances Claude’s responses with educational insights about:
  • Implementation choices and trade-offs
  • Codebase patterns and conventions
  • Design decisions specific to your project
  • Why certain approaches were chosen

​
Installation

This plugin is bundled with Claude Code. To enable it: 
/plugin enable explanatory-output-style

​
How It Works

The plugin uses a SessionStart hook to inject additional context at the beginning of every session. This instructs Claude to provide brief educational explanations before and after writing code.

​
Insight Format

Educational insights appear in this format: 
★ Insight ─────────────────────────────────────
[2-3 key educational points]
─────────────────────────────────────────────────

​
What You’ll Learn

Implementation Choices

Why specific approaches were selected for your codebase

Patterns & Conventions

Existing patterns in your code and how they’re applied

Trade-offs

Design decisions and their implications

Codebase-Specific Details

Insights tailored to your project, not generic advice

​
Example Interaction

When you ask Claude to implement a feature, you’ll see: 
"Add authentication to the API"
Claude’s response: 
★ Insight ─────────────────────────────────────
• Your codebase uses JWT tokens stored in httpOnly cookies
  for security (prevents XSS token theft)
• The middleware pattern in auth/middleware.ts follows your
  existing approach for request preprocessing
• Session timeout is set to 24h to match your user
  experience requirements from config/app.ts
─────────────────────────────────────────────────

Let me implement the authentication middleware...

[Code implementation]

★ Insight ─────────────────────────────────────
• Added rate limiting on login endpoint to prevent brute
  force attacks (consistent with your API security policy)
• Error messages intentionally vague ("Invalid credentials")
  to avoid username enumeration attacks
─────────────────────────────────────────────────

​
Migration from Output Styles

This plugin replaces the deprecated outputStyle setting:

​
Before (deprecated)

{
  "outputStyle": "Explanatory"
}

​
After (current)

/plugin enable explanatory-output-style

​
SessionStart Hook Pattern

This plugin demonstrates the SessionStart hook pattern, which is:
  • More flexible than the old output styles
  • Distributable through plugins
  • Similar to CLAUDE.md but for runtime behavior
For tasks beyond software development, use subagents instead of SessionStart hooks. Subagents change the system prompt, while SessionStart hooks add to it.

​
Managing the Plugin

Keep installed but turn off temporarily:
/plugin disable explanatory-output-style

​
Token Usage

The plugin adds approximately:
  • ~200 tokens to system prompt (per session)
  • ~50-150 tokens per insight shown
For a typical 20-turn session with 5 insights: ~450-950 additional tokens

​
Details

Name: explanatory-output-styleType: SessionStart HookAuthor: Dickson Tsai ([email protected])Version: 1.0.0Hook File: hooks-handlers/session-start.sh

Learning Output Style

Interactive learning with code contributions

Plugin Development

Create your own SessionStart hooks

Build docs developers (and LLMs) love

Get started for freeTalk to us