Root cause pareto
Skill gulmezeren2-byte/industrial-engineering-ai-skills/skills/root-cause-pareto
An operations-analysis method for AI agents: skills, role agents, data-hygiene rules and artifact templates - OTIF audits, forecast reviews, ABC-XYZ, safety stock, Pareto discipline, reporting contracts. Pitfalls included.
npx -y skills add gulmezeren2-byte/industrial-engineering-ai-skills --skill root-cause-paretoAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 23 days oldThe repository was created 23 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Build a decision-grade Pareto for downtime, defects, complaints or delays - with unit-of-measure discipline, category hygiene, exposure normalization and a follow-up metric. Use when the user mentions pareto analysis, kök neden, duruş analizi, downtime, arıza analizi, defect ranking, şikayet analizi, 80/20 analysis. Differentiator - treats the Pareto as the start of a causal investigation, not a bar chart deliverable.
SKILL.md
2.9 KB, as published. Nobody here has run it
Root-Cause Pareto
A Pareto chart is easy; a Pareto that survives challenge in a management meeting is not. The difference is four disciplines applied before the chart exists.
Workflow
- Pick the unit of measure deliberately. Occurrences, minutes, or money - choose the one closest to the pain being managed and say why. Ranking breakdowns by count when one category costs 10x more minutes per event points the team at the wrong problem. When in doubt, show count and impact side by side.
- Enforce category hygiene before counting:
- Categories must sit at one granularity level (no "Mechanical failure" next to "Sensor S-114 misaligned")
- "Other/Miscellaneous" must stay under ~15% of the total; if it is bigger, the categorization failed - split it before proceeding
- Merge synonyms and near-duplicates (free-text logs always contain them; list the merges made)
- Normalize by exposure before comparing. Line A with 3 shifts will "lead" any raw ranking against Line B with 1 shift. Divide by machine-hours, orders, or units produced when comparing across lines, shifts, or periods - state the exposure base used.
- Check stability. A Pareto from one bad week is an anecdote. Compare the ranking across at least two comparable periods; only categories that stay on top deserve investment. Note rank changes.
- Rank, plot, and go one level deeper on the #1 category. Break the top category into its own sub-Pareto or apply 5-why prompts to its most frequent instances. The actionable cause is usually one level below the headline category.
- Define the counter-metric. Before recommending an action, state which number should move, by roughly how much, and when to re-measure. A Pareto without a follow-up measurement is decoration.
Pitfalls to check explicitly
- Unit mismatch (count vs duration vs cost) silently reordering priorities
- Unnormalized cross-line comparisons
- "Other" as the tallest bar
- Categories mixing symptoms ("stopped") with causes ("no material")
- Acting on unstable rankings from short windows
Output format
- One sentence: unit of measure, exposure base, period, and total impact covered
- The ranked table (category, impact, share, cumulative) - chart optional, table mandatory
- Sub-analysis of the top category with candidate root causes
- Recommended action + counter-metric + re-measure date
- Data notes: merges performed, rows excluded, Other%