DS-STAR: Multi-Agent Data Science Framework

Overview

DS-STAR (Data Science - Structured Thought and Action) is an intelligent multi-agent framework adapted for Claude Code that automates data science workflows through specialized AI agents. Originally designed for Google's Gemini models, this skill adapts DS-STAR to leverage Claude's model family (Haiku, Sonnet, Opus) for cost-efficient, high-quality data analysis.

What It Does

DS-STAR orchestrates seven specialized agents that work collaboratively to:

1. Analyze data files (structured/unstructured)
2. Plan analysis strategies iteratively
3. Generate Python code to execute plans
4. Execute code with automatic debugging
5. Verify results against the query
6. Route refinement decisions
7. Finalize formatted outputs

Key Innovation: Multi-model optimization - route exploration tasks to Haiku, complex reasoning to Sonnet, and critical analysis to Opus for optimal cost/performance balance.