Can I use this for specific features instead of the whole repo?

Absolutely. You can provide a specific topic or feature as an argument (e.g., /ork:explore authentication) to focus the agents on a specific subset of the code.

How does the code health assessment work?

The skill uses a standardized rubric to evaluate the repository on a scale of 0-10, looking at factors like complexity, documentation, and adherence to design patterns.

Does this skill require specific MCP servers?

Yes, it requires a memory MCP server to store and retrieve context about the codebase's knowledge graph and previous discoveries for deeper analysis.

What is the benefit of using parallel agents for exploration?

Parallel agents allow Claude to analyze different aspects of the code—such as backend architecture, data flow, and code quality—simultaneously, providing a much faster and more holistic view than a single-threaded pass.

What is the difference between Task tool and Agent Teams mode?

Task tool mode uses a star pattern where a lead synthesizes findings at the end, while Agent Teams mode allows explorers to share discoveries in real-time for better cross-referencing.

Deep Codebase Explorer

Name: Deep Codebase Explorer
Author: yonatangross

byyonatangross

•

Productivity & Workflow

Conducts multi-angle codebase analysis and architectural discovery using parallel AI agents to map dependencies, health, and data flows.

Deep Codebase Explorer is a sophisticated orchestration skill for Claude Code designed to provide a comprehensive understanding of complex repositories. It utilizes a multi-agent strategy to simultaneously analyze code structure, data flow, architectural patterns, and code health. Whether you are onboarding onto a new project or tracing complex logic through a legacy system, this skill automates the discovery process, generating detailed reports that highlight dependency hotspots and product context while scoring code quality across the entire codebase using standardized rubrics.

Key Features

01Dependency Hotspot Mapping: Identifies and visualizes tight coupling and architectural bottlenecks within the system.

02Parallel Multi-Agent Analysis: Deploys up to 5 specialized agents to examine code from different perspectives simultaneously.

03Automated Code Health Scoring: Evaluates repository quality on a 0-10 scale based on standardized scoring criteria.

04Interactive Intent Verification: Clarifies exploration goals to optimize workflows for data flow, architecture, or quick searches.

0592 GitHub stars

06Comprehensive Exploration Reports: Generates actionable documentation covering findings, file locations, and design patterns.

Use Cases

01Onboarding onto a massive legacy repository to quickly understand core architecture and file locations.

02Tracing the end-to-end data flow of a specific feature, such as authentication or payment processing.

03Performing a codebase health audit to identify technical debt and high-risk dependency areas before a major refactor.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit explore

For use in Claude.ai and ChatGPT

Download Skill