Claude Code Optimizer Agent

Research Claude Code best practices, MCP patterns, performance optimization, context management, and agentic techniques with automated URL tracking and timestamped research reports.

Features

• Planning Step: Pre-research planning via /plan-claude-code-optimization for scoped, efficient research
• Systematic Research: Prioritizes changelog, HN, anthropic.com/engineering, and GitHub
• URL Tracking: Scored database (1-10) of all research sources
• Automated Reports: Timestamped summaries with TL;DR, findings, actions
• Thinking Modes: ultrathink (complex), think harder (compare), think (quick)

Installation

prpm install @t0dorakis/claude-code-optimizer

Prerequisites

Create required directories before first use:

mkdir -p .claude/research-logs/reports

Usage

Recommended: Plan First

/plan-claude-code-optimization optimize context usage

Presents plan, asks confirmation, then invokes optimizer.

Direct Agent Call

/agent claude-code-optimizer

Example Prompts

• "Research MCP server best practices"
• "Find context management strategies"
• "Compare agentic architectures"
• "Check Claude Code changelog for recent changes"

Directory Structure

.claude/
└── research-logs/
    ├── claude-code-sources.md      # Scored URL database
    └── reports/
        └── YYYY-MM-DD-HHMM-{topic}.md

URL Scoring

9-10: Official docs | 7-8: Quality tutorial | 5-6: Useful discussion | 1-4: Limited value

Report Format

# {topic} | {timestamp}

## TL;DR
{1-2 sentences, actionable}

## Findings
- **{Category}:** {Finding} [[score] source]

## Actions
- Now: {immediate}
- Later: {strategic}

## Sources
Top: [9] url | [8] url

Configuration

Model

Uses Opus by default for deep research quality.

Priority Sources

1. Claude Code changelog/releases
2. Hacker News ("Claude Code" OR "Anthropic")
3. anthropic.com/engineering
4. GitHub anthropics/claude-code

MCP Integration

Uses Context7 for framework documentation lookup.

Contributing

CONTRIBUTING.md | Open Issue

License

MIT - t0dorakis