repeated-failure-learning

name	repeated-failure-learning
description	Prove whether failures share one recurring root cause.

Repeated Failure Learning

Trigger

• Same root cause or failed approach occurs twice.
• User reports recurrence.
• Unrelated failures != recurrence.
• Attempted fix recurs → prior cause = unproven → return $diagnosing-bugs; extending workaround = forbidden.

Ownership

• Diagnosis = $diagnosing-bugs.
• Regression strength = $test-quality.
• Gate contract + commands/hooks/CI = $deterministic-checks.
• Active Feature Loop → record recurrence evidence → hand packet to $he-learn; lifecycle continues.
• Block delivery only when recurrence proves continued work risks security/privacy/accessibility/data integrity/data loss/irreversible action.
• This skill owns recurrence + root-cause evidence only; prevention selection = $he-learn.

Evidence

Field	Required proof
Signature	Stable error/symptom + affected boundary
Attempts	Change + result for each comparable attempt
Recurrence	Same root cause or failed approach ≥2 times
Cause	Root owner + why prior action failed
Success	Correcting change + passing proof
Exposure	Callers/data/routes/tests/docs/config affected

Complete

• Recurrence proven by comparable evidence.
• Root cause + successful correction identified.
• Evidence packet = signature + attempts + cause + correction + exposure.
• Active lifecycle → $he-learn packet handed off without routine source pause; otherwise evidence packet reported.

Report = PASS | CONCERNS | FAIL → recurrence proof → root cause → correction → exposure → handoff/gap.