skills.rest
Buckeyes22
Buckeyes22COMMUNITY
Explore all skills by this author →
GitHub

vibe-tuning

Diagnose AI output failures and generate approved enforcement artifacts.

Updated Aug 27, 2026
Search Tags:#hooks#memory#prompts#enforcement#postmortem#vibe-tuning#ai-calibration
One-click install
npx skills add https://github.com/Buckeyes22/weather-app --skill vibe-tuning
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: vibe-tuning
Source: https://github.com/Buckeyes22/weather-app/tree/main/docs/corpora/vibe-tuning
Command: npx skills add https://github.com/Buckeyes22/weather-app --skill vibe-tuning

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

vibe-tuning provides a structured, systematic way to diagnose and fix wrong AI results, ensuring mistakes don't repeat across sessions.

Core Features & Use Cases

  • 6-Step postmortem loop: CATCH, DIAGNOSE, ROOT CAUSE, FIX, PROPOSE SAVE, PROPOSE ENFORCE, all in dialog with user approvals.
  • Root-cause taxonomy and actionable fixes (Rule/Tool/Config/Education/Process) and the ability to propose and generate enforcement artifacts (hooks, CLAUDE.md rules, checklists).
  • Works across prompts, tools, and runs; generates persistent feedback that improves future sessions.

Quick Start

Initiate a vibe-tuning session by describing a wrong AI result and asking the AI to diagnose and propose a fix, with explicit user approval before saving or enforcing.

Frequently Asked Questions about vibe-tuning

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

FAQPage Schema
How do I stop AI from repeating the same mistakes across different sessions?▼

To stop AI from repeating mistakes across sessions, you need a structured postmortem process that diagnoses root causes and generates persistent enforcement artifacts like hooks or CLAUDE.md rules. This ensures corrections are applied to future interactions.

What is the best way to enforce rules after an AI generates an unsafe response?▼

The best way to enforce rules after an unsafe AI response is to run a diagnostic postmortem that proposes enforceable fixes, generating artifacts like hooks, CLAUDE.md rules, or checklists to prevent future occurrences.

How do I diagnose the root cause of wrong AI outputs in my prompts?▼

You can diagnose the root cause of wrong AI outputs by running a structured postmortem loop that categorizes the issue and proposes actionable fixes across rules, tools, configs, education, or processes.

Can I automate AI calibration without manually writing new prompts every time?▼

Yes, you can automate AI calibration by using a dialog-driven workflow that diagnoses wrong results and automatically generates persistent memory-backed improvements like hooks or rules, requiring only your approval before saving.

Does vibe-tuning generate persistent memory artifacts for future AI runs?▼

Yes, vibe-tuning generates persistent memory artifacts for future AI runs by outputting enforceable fixes like hooks, CLAUDE.md rules, and checklists after a structured postmortem diagnoses the root cause of an issue.

When should I use a postmortem workflow for AI responses?▼

You should use a postmortem workflow for AI responses when outputs are wrong, ambiguous, or unsafe across sessions, prompts, and tools, requiring a structured diagnosis to prevent the error from recurring.

Related Skills

Explore More Software Engineering Skills→
RintaroMasaokaRintaroMasaoka

improve

Diagnose root causes in agent prompts and suggest universal improvements.

Community
Advanced
alexeyvalexeyv
4

panout-help

Configure kitchen profiles, calibrate sensors, and monitor cooking sessions.

Community
Advanced
h4v1dh4v1d

systematic-debugging

Automate root-cause analysis before fixes using a four-phase debugging workflow.

Community
Advanced
ollieb89ollieb89

llm-evaluation

Evaluate LLM outputs with automated metrics and LLM-as-Judge workflows.

Community
Advanced