skills.rest
Alchemication
AlchemicationCOMMUNITY
Explore all skills by this author →
GitHub

feedback-triage

Diagnose incorrect AI coaching outputs by analyzing feedback traces and assigning failures to pipeline stages.

1|1|Updated Mar 16, 2026
Search Tags:#ai evaluation#trace analysis#prompt debugging#llm debugging#feedback triage#health coaching
One-click install
npx skills add https://github.com/Alchemication/zdrowskit --skill feedback-triage-alchemication
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: feedback-triage
Source: https://github.com/Alchemication/zdrowskit/tree/main/.agents/skills/feedback-triage
Command: npx skills add https://github.com/Alchemication/zdrowskit --skill feedback-triage-alchemication

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps identify why an AI coaching response, nudge, or insight failed by tracing the generation pipeline and separating model, prompt, data, and verification issues.

Core Features & Use Cases

  • LLM Trace Analysis: Inspect feedback records, call traces, tool results, verification stages, and rewrites to locate failure points.
  • Bug Ownership Assignment: Determine whether problems originate from source generation, verifier behavior, rewriting, prompt assembly, or stale health data.
  • Use Case: When a user reports a misleading health coach nudge, use this Skill to inspect the trace, validate the underlying data, and prepare a focused handoff for evaluation.

Quick Start

Use the feedback-triage skill to investigate the recent thumbs-down feedback and identify the failing stage.

Frequently Asked Questions about feedback-triage

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

FAQPage Schema
How do I debug bad AI coaching outputs using LLM trace analysis?▼

Identify why an AI health coach nudge failed by tracing the generation pipeline and separating model, prompt, data, and verification issues. It inspects feedback logs and assigns failures to stages like source generation, rewriting, or stale health data.

How do I assign bug ownership when a health assistant nudge provides incorrect feedback?▼

Assign bug ownership for incorrect nudges by inspecting LLM feedback traces and validating underlying health data. The process separates prompt assembly problems, verifier behavior issues, and stale data to pinpoint the exact failing pipeline stage.

What is the best way to diagnose incorrect LLM feedback in health coaching workflows?▼

The best way to diagnose incorrect LLM feedback is to use feedback triage to trace generation pipelines, inspect verification steps, and prepare a focused handoff for evaluation. This ensures reproducible regressions are identified across coaching workflows.

Do I need feedback logs and trace inspection tools to diagnose AI prompt failures?▼

Yes, you need access to feedback logs, trace inspection tools, and evaluation workflows to diagnose AI prompt failures. These inputs are required to reproduce regressions and analyze LLM traces for source generation and verification issues.

Why does my AI health coach nudge fail verification stages during prompt debugging?▼

Your AI health coach nudge fails verification stages due to verifier behavior issues, prompt assembly errors, or stale health data. LLM trace analysis inspects these verification stages and rewrites to locate the exact failure point.

Related Skills

Explore More Software Engineering Skills→
ouakarouakar

data-scientist

Document production AI/ML stack usage, LLM calls, pipelines, and cost drivers.

Community
Advanced
venetanjivenetanji
3

drama-video

Plan and execute a dialogue-driven video pipeline from per-shot prompts to final assembly.

Community
Advanced
jared-hawkins-youngjared-hawkins-young

meeting-recap-checkin

Extract action items and coaching notes from customer check-in transcripts.

Community
Advanced
faroukBakarifaroukBakari
5

debug-hypothesis

Organize hypothesis-driven debugging to isolate root causes of software failures.

Community
Advanced