Self Learning Loop

Use this skill between orbit rounds to make each subsequent round smarter. The loop learns which domains produce the most bugs, which fixes regress, and where to focus attack effort.

When to Use

• Before starting a new orbit round (R53 after R52)
• After completing a round to record findings
• When triaging to check if a finding is a regression

Workflow

1. Record findings after a round

After R52 completes, record all findings with domain, severity, and fix status.

2. Generate learning report before R53

Query all prior findings. The report includes:

• Domain weights — domains with more critical/unfixed findings get higher priority
• Regression candidates — previously-fixed findings flagged for re-verification
• Recommended priorities — high/medium/low per domain based on evidence
• — how much data the loop has to make good recommendations (0.0 = no data, 1.0 = sufficient data)