What is the Code Discovery skill for Claude Code?

Code Discovery is a specialized skill that provides patterns for deep codebase research, helping Claude understand code history, trace complex implementations, and find existing precedents within a project.

What are 'precedents' in the context of this skill?

Precedents are existing code patterns or similar features already implemented in your project. Finding them ensures that new code remains consistent with the project's established architecture and style.

When should I use the Code Discovery pattern?

You should use it whenever you need to explore code you haven't seen before, find similar features to follow as a precedent, or provide concrete file references for a proposal.

How does this skill help with bug investigation?

It uses configurable git log timeframes and path-specific searches to identify exactly when and where changes were introduced, allowing for evidence-driven regression analysis.

Code Discovery

Name: Code Discovery
Author: flox

byflox

•

開発者ツール

Performs deep codebase research and investigation using systematic git analysis and evidence-based code reading strategies.

The Code Discovery skill provides a rigorous framework for navigating and understanding complex repositories, moving beyond simple keyword searches to deep structural analysis. It empowers developers to perform exhaustive git log analysis, trace implementations from public interfaces to core logic, and identify precedent patterns for new features. By prioritizing direct evidence over assumptions and utilizing standardized file:line referencing, it ensures that every design proposal or bug investigation is grounded in the actual implementation details of the codebase.

主な機能

01Configurable git log analysis for pinpointing regressions and change history.

02Pattern discovery tools to find similar features for architectural consistency.

03Standardized file:line referencing for verifiable and clickable findings.

04Systematic code reading strategies using context files as entry points.

051 GitHub stars

06Structured confidence levels to distinguish between hard facts and hypotheses.

ユースケース

01Onboarding to unfamiliar modules by tracing execution paths from entry points to handlers.

02Investigating suspected regressions by analyzing git history over specific timeframes.

03Conducting design research to find existing code patterns before implementing new features.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add flox/forge-plugin code-discovery

For use in Claude.ai and ChatGPT