What is the 'Iron Law' of this debugging skill?

The Iron Law dictates that no fixes may be proposed until a thorough Phase 1 Root Cause Investigation is completed and the source of the error is proven.

Does this skill require writing tests?

Yes, Phase 4 of the process requires creating a minimal failing test case to reproduce the bug before any fix is implemented, ensuring the solution is verified and permanent.

What happens if a bug persists after several fix attempts?

If three or more fixes fail, the skill signals an architectural red flag, prompting a discussion about fundamental design flaws rather than continuing to fix symptoms.

How does this skill handle multi-component systems?

It implements diagnostic instrumentation at every component boundary (e.g., API to Database or CI to Build) to log data entry and exit points, pinpointing exactly where the logic breaks.

Is this approach slower than just fixing the bug directly?

While it requires more initial investigation, it is significantly faster than 'guess-and-check' thrashing, which often leads to regressions and wasted hours of debugging.

Shipyard Systematic Debugging

Name: Shipyard Systematic Debugging
Author: lgbarn

bylgbarn

•

セキュリティとテスト

Enforces a disciplined four-phase debugging framework to identify root causes and eliminate inefficient guess-and-check fixing cycles.

Shipyard Systematic Debugging is a specialized Claude Code skill designed to eliminate the 'thrashing' often associated with complex software issues. By mandating a strict progression through Root Cause Investigation, Pattern Analysis, Hypothesis Testing, and Implementation, it ensures that Claude never applies a 'band-aid' fix without understanding the underlying failure. It is particularly powerful for multi-component systems and CI/CD pipelines, providing structured diagnostic instrumentation and architectural red-flag detection when simple patches fail to resolve persistent problems.

主な機能

01Strict four-phase debugging protocol to prevent premature or random code changes.

02Architectural health checks that trigger after three failed fix attempts to identify structural flaws.

0338 GitHub stars

04Integration with TDD principles requiring a failing test case before any implementation.

05Backward data flow tracing to isolate the origin of corrupted values in deep stacks.

06Multi-layer diagnostic instrumentation for tracing data across component boundaries.

ユースケース

01Resolving opaque errors in complex distributed systems where the point of failure is unknown.

02Providing structured evidence gathering during high-pressure production incidents or build failures.

03Debugging persistent issues that have resisted multiple previous 'quick fix' attempts.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lgbarn/shipyard shipyard-debugging

For use in Claude.ai and ChatGPT

Download Skill