When should Claude delegate a task to Gemini?

Delegation is recommended when the context exceeds 150,000 tokens, or when specific vision capabilities like complex OCR or PDF parsing are required.

Can I use different Gemini models with this skill?

Yes, the skill is model-agnostic. You can swap the underlying model by simply updating the GEMINI_MODEL environment variable without changing your logic.

What is the 'Validation Rule' in Gemini Analyst?

It implements a 'propose with Claude, validate with Gemini' pattern, using Gemini to provide an independent assessment of security reviews or design choices.

How does Gemini Analyst handle large codebases?

It utilizes the gemini_analyze_codebase tool, which is specifically designed to process and analyze repository content up to 1 million tokens.

Gemini Analyst

Name: Gemini Analyst
Author: talent-factory

bytalent-factory

•

Data Science & ML

Delegates high-context tasks and complex vision analysis from Claude to Gemini for optimized multi-model development workflows.

Gemini Analyst is a sophisticated bridge skill that empowers Claude to intelligently delegate tasks to Gemini models based on token budget, vision requirements, and validation needs. It solves the challenge of processing massive codebases by routing tasks exceeding 150,000 tokens to Gemini's expansive context window, while also leveraging Gemini's specific strengths in complex OCR and PDF parsing. By implementing a 'propose with Claude, validate with Gemini' pattern, it provides an independent layer of verification for critical architectural decisions and security reviews within a model-agnostic framework.

Key Features

01Standardized output attribution for transparent audit trails

02Advanced vision analysis for complex diagrams and multi-page PDFs

03Multi-model architectural validation and security cross-referencing

04Model-agnostic configuration via environment variables

053 GitHub stars

06Automatic token-based routing for codebases up to 1M tokens

Use Cases

01Validating complex system design choices with an independent model assessment

02Analyzing massive legacy repositories that exceed standard LLM context limits

03Extracting technical requirements from large architectural diagrams and documentation

Key Features

01Standardized output attribution for transparent audit trails

02Advanced vision analysis for complex diagrams and multi-page PDFs

03Multi-model architectural validation and security cross-referencing

04Model-agnostic configuration via environment variables

053 GitHub stars

06Automatic token-based routing for codebases up to 1M tokens

Use Cases

01Validating complex system design choices with an independent model assessment

02Analyzing massive legacy repositories that exceed standard LLM context limits

03Extracting technical requirements from large architectural diagrams and documentation