skills.rest
patricio0312rev
patricio0312revCOMMUNITY
Explore all skills by this author β†’
Website𝕏@patricio0312revGitHub

llm-debugger

Diagnose LLM output failures including hallucinations, format violations, and reasoning errors.

5|Updated Dec 31, 2025
Search Tags:#testing#debugging#llm#diagnosis#json-validation#prompt-fix
One-click install
npx skills add https://github.com/patricio0312rev/skillset --skill llm-debugger
Or copy as Structured Prompt for Agentβ–Ό
Please help me install this Agent Skill.
Skill: llm-debugger
Source: https://github.com/patricio0312rev/skillset/tree/main/templates/ai-engineering/llm-debugger
Command: npx skills add https://github.com/patricio0312rev/skillset --skill llm-debugger

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Diagnoses LLM output failures including hallucinations, formatting issues, and reasoning gaps, enabling root cause analysis and guided fixes.

Core Features & Use Cases

  • Root-cause classification of LLM failures (hallucination, format_violation, constraint_break, reasoning_error, tool_misuse, refusals, incomplete_output)
  • Prompt fixes and iterative testing to stabilize AI outputs
  • Tooling improvements and test-case generation for robust AI-assisted coding tasks
  • Real-world scenario: debugging a misbehaving code assistant and validating structured outputs

Quick Start

Start an interactive debugging session to identify and fix LLM output issues.

Frequently Asked Questions about llm-debugger

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

FAQPage Schema
How do I debug LLM output failures like hallucinations and format violations?β–Ό

Debug LLM output failures by classifying root causes such as hallucinations, format violations, and reasoning errors. This diagnostic process enables targeted prompt fixes and iterative testing to stabilize structured AI outputs.

What is the best way to fix JSON validation errors in LLM generated outputs?β–Ό

Fix JSON validation errors by diagnosing them as constraint breaks or format violations. Applying root-cause classification identifies the specific formatting failure, enabling actionable prompt adjustments for reproducible structured outputs.

How do I diagnose reasoning errors and tool misuse in AI-assisted coding tasks?β–Ό

Diagnose reasoning errors and tool misuse in AI-assisted coding by classifying the LLM's failure mode. This approach identifies whether the issue stems from reasoning gaps or incorrect tool usage to guide workflow improvements.

Can I generate test cases to validate structured LLM outputs for prompt design?β–Ό

You can generate test cases to validate structured LLM outputs during prompt design. This Skill supports test-case generation to ensure explicit constraints are satisfied and to stabilize outputs across iterative debugging sessions.

Why does my LLM output incomplete responses or refusals despite clear prompt instructions?β–Ό

Incomplete responses or refusals occur due to constraint breaks or reasoning errors. Diagnosing these specific LLM output failures reveals the root cause, allowing you to apply actionable prompt fixes for stable generation.

Related Skills

Explore More Software Engineering Skills→
michaelsvanbeekmichaelsvanbeek
1

anomaly-detection

Detect unusual patterns in time-series, logs, and metric data.

Community
Advanced
semicolon-devteamsemicolon-devteam

validate-test-cases

Validate test cases against issue acceptance criteria and generate verification reports.

Official
Advanced
Fadil369Fadil369

pptx-medical

Generate medical PowerPoint presentations from FHIR resources and lab results.

Community
Advanced
JhabbigJhabbig

llm-evaluation

Design evaluation frameworks for LLM applications with automated and human scoring.

Community
Advanced