How does this skill manage token usage?

It uses specialized tldr-explorer commands to analyze structure and metadata, achieving up to 95% token savings compared to reading raw files.

Can the output be used by other AI agents?

Yes, the --output handoff option creates a standardized YAML file designed to feed context directly into development tools like /build.

Is the Explore skill safe for production code?

Yes, the Explore skill is strictly read-only and does not modify any source code during the analysis process.

Can I target the exploration to a specific directory?

Absolutely. You can use the --focus option to target specific modules, directories, or concepts such as 'authentication' or 'API routes'.

What is the difference between Deep and Architecture depth?

Deep depth focuses on pattern documentation and thorough code research, while Architecture depth maps system boundaries, call graphs, and identifies circular dependencies.

Codebase Explorer

Name: Codebase Explorer
Author: parcadei

byparcadei

•

3,382

•

Learning & Documentation

Analyzes and maps internal codebases at multiple depths to provide rapid orientation and deep architectural insights.

The Explore skill is a meta-utility designed to automate the process of understanding unfamiliar or complex brownfield codebases. It offers three distinct depth levels—Quick, Deep, and Architecture—allowing developers to rapidly orient themselves, identify design patterns, or map system boundaries without polluting the context window. By leveraging token-efficient analysis and generating standardized handoff files, it seamlessly bridges the gap between initial discovery and active development workflows.

Key Features

01Token-efficient analysis using tldr-explorer integration

023,382 GitHub stars

03Automated architectural layer and dependency mapping

04Multi-depth exploration modes (Quick, Deep, and Architecture)

05Interactive question flow for guided codebase discovery

06Standardized documentation and handoff generation for agent orchestration

Use Cases

01Mapping architectural layers and circular dependencies before refactoring

02Generating structured context files for implementation-focused AI agents

03Onboarding onto a large, unfamiliar project or legacy codebase

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add parcadei/continuous-claude-v3 explore

For use in Claude.ai and ChatGPT

Download Skill