How does the cost optimization work in this skill?

It routes simple tasks like data inspection to the low-cost Claude Haiku model, while reserving complex reasoning and coding for Sonnet or Opus, resulting in significant cost savings compared to single-model approaches.

What kind of files can I analyze with this skill?

DS-STAR supports a wide variety of data formats including CSV, JSON, and plain text, and can process both structured datasets and unstructured information.

What is DS-STAR for Claude Code?

DS-STAR is a multi-agent framework that uses a 'Structured Thought and Action' architecture to automate data science tasks like exploratory analysis and code execution directly within your development environment.

Does DS-STAR handle code errors automatically?

Yes, the framework includes a dedicated Debugger agent that identifies execution errors and automatically generates fixes through iterative refinement rounds.

Are the analysis results reproducible?

Absolutely. Every run creates a structured directory containing the exact prompts used, the generated Python scripts, and comprehensive execution logs for full transparency.

DS-STAR Data Science Framework

Name: DS-STAR Data Science Framework
Author: Token-Eater

byToken-Eater

Data Science & ML

Automates complex data science workflows using a multi-agent architecture and optimized model routing for efficient, iterative data analysis.

About

DS-STAR (Data Science - Structured Thought and Action) is a sophisticated multi-agent framework designed to transform raw data into actionable insights within Claude Code. By orchestrating seven specialized agents—ranging from data analyzers to Python coders and verifiers—it manages the entire data science lifecycle including planning, execution, and iterative debugging. The framework optimizes performance and cost by dynamically routing tasks to the most suitable Claude model (Haiku, Sonnet, or Opus), providing a reproducible and highly efficient alternative to manual exploratory data analysis.

Key Features

Comprehensive artifact generation that saves prompts, code, and logs for full research reproducibility
Cost-optimized model routing across Haiku, Sonnet, and Opus to reduce analysis expenses by up to 60%
Iterative refinement cycles with automated debugging and verification gates for high-accuracy results
Multi-agent orchestration using specialized roles for data analysis, strategy planning, and coding
Native support for diverse data formats including CSV, JSON, and unstructured text files
0 GitHub stars

Use Cases

Automating exploratory data analysis (EDA) and generating statistical summaries for new datasets
Performing cost-effective data processing workflows using mixed-model logic for large datasets
Executing iterative time-series or correlation analysis with automated Python code generation and debugging

About

Key Features

Comprehensive artifact generation that saves prompts, code, and logs for full research reproducibility
Cost-optimized model routing across Haiku, Sonnet, and Opus to reduce analysis expenses by up to 60%
Iterative refinement cycles with automated debugging and verification gates for high-accuracy results
Multi-agent orchestration using specialized roles for data analysis, strategy planning, and coding
Native support for diverse data formats including CSV, JSON, and unstructured text files
0 GitHub stars

Use Cases

Automating exploratory data analysis (EDA) and generating statistical summaries for new datasets
Performing cost-effective data processing workflows using mixed-model logic for large datasets
Executing iterative time-series or correlation analysis with automated Python code generation and debugging