wasabeef/claude-code-cookbook/token-efficient

Compress responses to reduce context usage by 30-50%

Reduces AI response context usage by 30-50% through compression efficiency mode.

Compress responses to reduce context usage by 30-50%

Token Efficiency Mode leverages visual symbols and abbreviation systems to compress Claude's responses. Generated code quality and content remain unchanged. Only the explanation method changes.

Usage

# Enable mode
"Respond in Token Efficiency Mode"
"--uc mode"
"Concise mode"

How It Works

1. Symbol System

Logic & Flow

Status & Progress

Technical Domains

2. Abbreviation System

System & Architecture

• cfg → configuration
• impl → implementation
• arch → architecture
• perf → performance
• ops → operations
• env → environment

Development Process

• req → requirements
• deps → dependencies
• val → validation
• auth → authentication
• docs → documentation
• std → standards

Quality & Analysis

• qual → quality
• sec → security
• err → error
• rec → recovery
• sev → severity
• opt → optimization

Example Comparisons

Example 1: Error Report

Normal Mode (85 chars)

Security vulnerability found in the user validation function at line 45 of the auth system.

Token Efficient (39 chars)

auth.js:45 → 🛡️ sec vuln in user val()

Example 2: Build Status

Normal Mode (112 chars)

Build process completed successfully. Tests are currently running, followed by deployment.

Token Efficient (35 chars)

build ✅ » test 🔄 » deploy ⏳

Example 3: Performance Analysis

Normal Mode (95 chars)

Performance analysis revealed slow processing due to O(n²) algorithm complexity.

Token Efficient (42 chars)

⚡ perf: slow ∵ O(n²) → optimize to O(n)

Use Cases

✅ Effective Scenarios

• Long debugging sessions: Efficiently maintaining history
• Large code reviews: Concise analysis of many files
• CI/CD monitoring: Real-time status updates
• Project progress reports: Overview of multiple task states
• Error tracking: Visual representation of problem chains

❌ Scenarios to Avoid

• Explanations for beginners
• Detailed documentation creation
• Initial requirements definition
• Communication with non-technical stakeholders

Implementation Examples

Debugging Session

[14:23] breakpoint → vars: {user: null, token: expired}
[14:24] step → auth.validate() ❌
[14:25] check → token.exp < Date.now() ∴ expired
[14:26] fix → refresh() → ✅
[14:27] continue → main flow 🔄

File Analysis Results

/src/auth/: 🛡️ issues × 3
/src/api/: ⚡ bottleneck in handler()
/src/db/: ✅ clean
/src/utils/: ⚠️ deprecated methods
/tests/: 🧪 coverage 78%

Project Status

Frontend: 🎨 ✅ 100%
Backend: ⚙️ 🔄 75%
Database: 🗄️ ✅ migrated
Tests: 🧪 ⚠️ 68% (target: 80%)
Deploy: 📦 ⏳ scheduled
Security: 🛡️ 🚨 1 critical

Configuration Options

// Compression levels
--uc; // Ultra Compressed: Maximum compression
--mc; // Moderate Compressed: Medium compression
--lc; // Light Compressed: Light compression
// Domain-specific
--dev; // Development-focused compression
--ops; // Operations-focused compression
--sec; // Security-focused compression

Benefits

1. Context saving: 30-50% token reduction
2. Visual understanding: Intuitive grasp through symbols
3. Information density: More information in the same space
4. History retention: Maintain longer conversation history
5. Pattern recognition: Easier problem detection through visual patterns

Notes

• This mode only changes AI response style
• Code quality remains unchanged
• Can switch with "explain in normal mode" as needed
• Normal mode recommended for beginners and non-technical users

Command Examples

# Enable
"Token Efficient Mode on"
"Respond concisely"
"Analyze with --uc"
 
# Disable
"Return to normal mode"
"Explain in detail"
"Token Efficient Mode off"

Implementation Impact

💡 Pro Tip: For long work sessions, start with normal mode to build understanding, then switch to Token Efficient Mode to optimize efficiency and context retention.