How does Kosmos X-Ray save tokens?

It uses AST parsing to extract only the structure (classes, methods, signatures) of a file, omitting the internal implementation code which often accounts for the majority of a file's token count.

Does it support visual diagrams?

Yes, it can generate Mermaid-formatted output for its dependency graph tool, which can be rendered in Markdown viewers to visualize system architecture.

Can it handle Pydantic models?

Yes, the enhanced skeleton tool explicitly captures Pydantic and dataclass fields, ensuring data structures are visible even when the logic implementation is hidden.

What is a 'context hazard'?

A context hazard refers to exceptionally large files that could consume an entire AI context window; the mapper tool identifies these early so you can use skeleton views instead.

Kosmos X-Ray

Name: Kosmos X-Ray
Author: jimmc414

byjimmc414

•

324

•

데이터 과학 및 ML

Explores and maps complex codebases using AST analysis to reduce token usage by 95% while maintaining structural visibility.

Kosmos X-Ray is a specialized diagnostic tool designed for high-efficiency codebase exploration within limited context windows. By leveraging Abstract Syntax Tree (AST) parsing, it extracts critical structural elements—such as Pydantic fields, decorators, and class hierarchies—without loading full implementation details. This skill is essential for developers working with large-scale projects like the Kosmos AI Scientist, enabling rapid onboarding, architectural visualization via Mermaid diagrams, and precise context management for AI-driven coding tasks.

주요 기능

01Directory mapping with token budget estimates and hazard detection

02324 GitHub stars

03Automated Mermaid diagram generation for architectural dependency graphs

04Priority-based filtering to focus on critical modules and core interfaces

05Support for Pydantic/dataclass field visibility and decorator mapping

06AST-based skeleton extraction for Python files with ~95% token savings

사용 사례

01Rapidly onboarding to complex AI research codebases without exceeding context limits

02Generating architectural documentation and dependency maps for large Python projects

03Identifying large files and context hazards before performing deep-dive code analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jimmc414/kosmos kosmos-xray

For use in Claude.ai and ChatGPT

Download Skill