Hold quality in control and stop defects recurring: stand up SPC reading special vs common cause correctly, structure a nonconformance through containment into a real CAPA, build inspection/control plans and FMEA, and set the supplier-quality bar — prevention over detection over scrap.
Installs into .claude/skills of the current project.
Are you the author of Capa And Spc?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/mcorbett51090-capa-and-spc)
---
name: capa-and-spc
description: "Hold quality in control and stop defects recurring: stand up SPC reading special vs common cause correctly, structure a nonconformance through containment into a real CAPA, build inspection/control plans and FMEA, and set the supplier-quality bar — prevention over detection over scrap."
---
# CAPA & SPC
## Special cause vs common cause first
SPC's first question: is the signal special or common cause? Don't tamper with common-cause noise (adjusting a stable process adds variation); don't ignore a special-cause signal. Control limits come from the voice of the *process* (the data), never from the spec (the voice of the customer) — conflating them is the most common SPC error. Pick the right chart (X̄-R, I-MR, p/np/c/u) and apply the run rules.
## NCR → containment → CAPA
A defect is a process signal, not just a part to scrap. Containment stops the bleeding; the CAPA fixes the cause. Separate corrective action (fix this batch) from preventive action (remove the root cause), run the root-cause analysis (5-Whys / fishbone / is–is-not), verify effectiveness, and update the control plan/FMEA. Containment alone is closing nothing.
## Prevention beats detection beats scrap
An FMEA (severity × occurrence × detection → action priority) or a poka-yoke that designs the failure out beats source detection, which beats catching defects at final inspection. Final inspection is the weakest, latest, leakiest control — push quality upstream. An FMEA that never changes a control is paperwork.
## Cpk is not a pass/fail stamp
A capability index on an *unstable* process is meaningless. Check stability first, state the sample size and measurement-system basis, then read capability. Route the Gage R&R / MSA and the capability-study math to `applied-statistics`.
## Supplier quality & no silent sign-off
Set the incoming-quality bar and PPAP expectations; route the sourcing decision to `procurement-sourcing`. A regulated CAPA closure, deviation disposition, or recall call is drafted here and escalated to the accountable human — never auto-closed.
## Output
A structured CAPA (containment → root cause → corrective + preventive → effectiveness check → control-plan/FMEA update) or an SPC setup (right chart, process-derived limits, run rules, special-vs-common-cause calls). Hand the inferential rigor to `applied-statistics` and designing-the-cause-out to `process-improvement`.
