repo-analyze

Analyze a code repository using Gemini CLI collaboration to leverage Gemini's large context window and Claude's detailed analysis capabilities.

Analysis Type: $ARGUMENTS (project, gitignored)

1. Environment Setup:

   • Verify Gemini CLI is available (npm install -g @google/gemini-cli if missing)
   • Validate repository path exists
   • Create output directory structure for results
   • Record analysis metadata (timestamp, tools, versions)

2. Gemini Global Analysis (Phase 1):

   • IMPORTANT: executing each step in this phase by using gemini -p
   • Execute comprehensive architecture scan using gemini -p
   • Analyze project structure and organization
   • Identify technology stack and dependencies
   • Map core components and their relationships
   • Detect circular dependencies and coupling issues
   • Generate JSON output for structured analysis

3. Detailed Code Analysis (Phase 2):

   • Review Gemini's global analysis results
   • Perform deep code quality assessment:
     • Cyclomatic complexity evaluation
     • Cognitive complexity and readability
     • Code duplication detection
     • Naming conventions and consistency
   • Identify performance bottlenecks:
     • Algorithmic complexity issues
     • N+1 query problems
     • Memory leak risks
     • Concurrency and race conditions
   • Security vulnerability scanning:
     • Input validation gaps (SQL injection, XSS, path traversal)
     • Authentication and authorization issues
     • Hardcoded credentials detection
     • Sensitive data exposure risks

4. Documentation Generation (Phase 3):

   • Create API documentation with usage examples
   • Generate architecture documentation with Mermaid diagrams
   • Produce developer guide covering:
     • Environment setup
     • Project structure explanation
     • Development workflow
     • Common tasks and debugging
     • Performance optimization tips

5. Report Compilation (Phase 4):

   • Generate comprehensive analysis report including:
     • Executive summary with key findings
     • Code quality metrics table
     • Security assessment with prioritized issues
     • Performance analysis and recommendations
     • Action plan (P0: critical, P1: short-term, P2: long-term)
   • Create interactive HTML dashboard with visualizations
   • Organize all outputs in structured directory

6. Advanced Options:

   • Batch processing mode for large repositories (>100k files)
   • Incremental analysis for changed files only
   • Custom analysis type combinations

Output structure saved to timestamped directory with subdirectories: raw_data/, reports/, documentation/, metrics/

Focus on actionable insights and prioritize findings by severity. Ask for clarification if repository scope or analysis depth is unclear.