Passed 13 of 20 test cases with the skill, and 6 of 20 without it.
Without skill32%
With skill64%
Checks it passed that the model alone missed
Captures the explicit delta or comparative improvement from prior executions
Explicitly tags the framework-level configuration issue as a meta-improvement
Executes a backward chain validation step from root causes back to the initial symptoms
and 6 more in the full report
7 fixed7 passed either way8 still failing0 broke
Pass-rate lift
+32 pp
Tokens
+35%
Agent turns
0%
Response time
+14%
Model gemini-3.6-flashMeasured 8/16/2026v114 of 22 cases passed with skill · 20 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
During our phase 3.5 retro, we realized that the nWave framework runner configuration lacked support for dynamic timeout limits, causing agent step failures. How should this finding be categorized?
What was checked, without and with the skill
Checked
Without skill
With skill
Explicitly tags the framework-level configuration issue as a meta-improvement
Missed
Passed
case-14 · gemini-3.6-flash · 8/16/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md58 linessingle file
NW-ROOT-WHY: Toyota 5 Whys Root Cause Analysis
Wave: CROSS_WAVE
Agent: Rex (nw-troubleshooter)
Overview
Systematic root cause analysis using Toyota's 5 Whys with multi-causal investigation and evidence-based validation. Investigates multiple cause branches at each level|validates solutions against all identified root causes.
Agent Invocation
@nw-troubleshooter
Execute *investigate-root-cause for {problem-statement}.
Configuration:
investigation_depth: 5
multi_causal: true
evidence_required: true
Usage: DELIVER Wave Retrospective (Phase 3.5)
When invoked as part of /nw-deliver Phase 3.5, execute in order:
Gather Inputs — Read the evolution document, mutation results, git log, roadmap.json, and execution-log.json. Gate: all available inputs loaded.
Analyze What Worked Well — Identify practices that succeeded and document why — preserve these. Gate: at least one practice documented with rationale.
Analyze What Improved — Identify what worked better than before and why — record as reinforcements. Gate: delta from prior execution captured.
Analyze What Worked Badly — Apply 5 Whys root cause analysis to failures; produce actionable fix per root cause. Gate: each failure has a root cause and fix.
Analyze What Regressed — Apply 5 Whys to anything worse than before; produce prevention action. Gate: each regression has a root cause and prevention action.
Tag Meta-Improvements — Mark any items requiring nWave framework changes as meta-improvements. Gate: all framework-level issues flagged.
Append Retrospective — Write retrospective section to evolution document. If clean execution (no skips, no failures, no tooling issues), generate brief summary only. Gate: evolution document updated.