repomix-package-builder
Create targeted repomix XML packages for AI code analysis. Suggests templates (metadata-flow, code-review, architecture, context) and file patterns based on task objectives. See reference.md for detailed workflows and patterns.
When & Why to Use This Skill
The Repomix Package Builder is a specialized Claude skill designed to streamline AI-driven code analysis by generating targeted XML packages. It optimizes context for Large Language Models (LLMs) by selecting relevant files based on specific objectives—such as architectural reviews, debugging, or documentation—ensuring efficient token usage and high-quality AI insights through structured repository bundling.
Use Cases
- AI-Powered Code Reviews: Automatically bundle implementation files, tests, and documentation to provide AI agents with a comprehensive yet concise context for code quality and standards compliance assessments.
- Architectural Analysis & Design: Generate packages focused on service boundaries, DDD layers, and system patterns to facilitate high-level architectural reviews and pattern identification.
- Efficient Debugging and Knowledge Transfer: Create 'context' packages that include only essential implementation files and docs to help AI or new team members troubleshoot specific issues without exceeding token limits.
- Cross-Service Integration Tracing: Trace data and metadata flows between services by packaging event schemas, API definitions, and flow logic for complex integration analysis.
| name | repomix-package-builder |
|---|---|
| description | Create targeted repomix XML packages for AI code analysis. Suggests templates (metadata-flow, code-review, architecture, context) and file patterns based on task objectives. See reference.md for detailed workflows and patterns. |
Repomix Package Builder
Quick-start tool for creating AI-ready code analysis packages saved to .claude/repomix_packages/.
When to Use
Activate when user needs to:
- Prepare code for external AI review or analysis
- Document cross-service integration flows (metadata, events, API calls)
- Conduct architectural analysis or pattern identification
- Create context packages for debugging or knowledge transfer
- Mentions: "repomix", "code package", "AI review", "context export"
Available Templates
- metadata-flow - Trace data/metadata between services via events and APIs
- code-review - Comprehensive implementation + tests + docs + standards compliance
- architecture - System design, service boundaries, DDD layers, patterns
- context - Quick overview with essential docs and key implementation files
Quick Workflow
- Understand task objective (review? debug? document? analyze?)
- Select matching template
- Identify scope (which services/components?)
- Build file include list (see
reference.mdfor patterns) - Generate filename:
repomix-{feature}-{purpose}.xml - Execute repomix command targeting
.claude/repomix_packages/
Reference Documentation
- Detailed Templates & Patterns: See
reference.mdin this directory - Real-World Examples: See
examples.mdin this directory - Slash Command: User can trigger
/repomixfor interactive workflow
Output Conventions
- Location:
.claude/repomix_packages/ - Naming:
repomix-{feature}-{purpose}.xml - Examples:
repomix-eng5-kafka-review.xml,repomix-metadata-population-task.xml
Token Budgets
- Quick context: 20-30K tokens (~7-10 files)
- Standard review: 60-80K tokens (~25-35 files)
- Deep analysis: 80-100K tokens (~30-40 files)
For comprehensive guidance on file selection, templates, and best practices, reference reference.md.