skills.rest
willyu1007
willyu1007COMMUNITY
Explore all skills by this author →
GitHub

debug-mode

Execute an evidence-driven debugging loop with instrumentation, approvals, and verification.

Updated Dec 22, 2025
Search Tags:#code quality#debugging#troubleshooting#verification#llm#instrumentation
One-click install
npx skills add https://github.com/willyu1007/Template-Skill-Basic --skill debug-mode-willyu1007
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: debug-mode
Source: https://github.com/willyu1007/Template-Skill-Basic/tree/main/.claude/skills/workflows/llm/debug-mode
Command: npx skills add https://github.com/willyu1007/Template-Skill-Basic --skill debug-mode-willyu1007

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill addresses the challenge of debugging AI-generated code or behavior by providing a structured, evidence-driven loop that ensures thorough analysis and verification before and after fixes.

Core Features & Use Cases

  • Instrumented Debugging: Automatically adds logging and instrumentation to code for better reproduction of issues.
  • Approval Gates: Requires explicit user approval at key stages, such as before applying a fix or after verification.
  • Automated Cleanup: Ensures that debug-specific instrumentation is removed once the debugging session is complete.
  • Use Case: When an LLM-generated function is producing inconsistent results, this Skill can be used to systematically instrument the function, reproduce the error, analyze the logs, apply a fix, and verify its effectiveness, all within a controlled process.

Quick Start

Initiate the debug-mode skill to analyze and resolve the intermittent errors in the provided code snippet.

Frequently Asked Questions about debug-mode

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug inconsistent results from AI-generated code?▼

Debug inconsistent AI-generated code using an evidence-driven debugging loop that instruments the function, reproduces the error, analyzes run_id-tagged logs, and applies a verified fix.

What is the best way to troubleshoot flaky issues in LLM-generated functions?▼

Troubleshoot flaky LLM-generated functions through a systematic process that hypothesizes root causes, adds temporary instrumentation to reproduce the behavior, and analyzes logs before applying fixes.

How do I add instrumentation to reproduce and analyze intermittent code errors?▼

Add instrumentation to reproduce intermittent errors by injecting logging into the target code, capturing execution data with run_id tags, and analyzing the output to isolate the root cause.

Does AI debugging require user approval before applying a code fix?▼

AI debugging requires explicit user approval at key stages, ensuring you review the analyzed logs and verify the proposed fix before changes are applied to the codebase.

How do I ensure debug instrumentation is removed after troubleshooting?▼

Ensure debug instrumentation is removed through an automated cleanup process that strips all debug-specific logging and temporary code additions once verification confirms the fix is effective.

When do I need multi-pass verification for code troubleshooting?▼

Multi-pass verification for code troubleshooting is needed when resolving flaky issues, ensuring the fix consistently resolves the error across multiple validation runs before cleaning up instrumentation.

Related Skills

Explore More Software Engineering Skills→
ponylangponylang
6

pony-debug

Diagnose non-trivial issues with checkpoint artifacts and hypothesis-driven experiments.

Official
Advanced
JNZaderJNZader

debug-mode

Automate hypothesis-driven debugging with tagged logs and automated cleanup.

Community
Advanced
lance-doniflance-donif

ops-inspector

Diagnose CloudBase resource health and analyze error logs across functions, CloudRun, and databases.

Community
Intermediate
The13thNodeThe13thNode
3

head-of-growth

Conduct weekly growth audits and design experiments for startup metrics.

Community
Advanced