Shadow Teacher
Autonomous gate calibration learning loop — governance & feedback dashboard
Pending Approval Queue
HIGH and CRITICAL stakes genes awaiting human review
| Gate / Marker | Type | Reason | Conf | Stakes | Created | Actions |
|---|---|---|---|---|---|---|
No genes pending approvalAll gene submissions have been processed or none require manual review. |
||||||
Feedback History
Recent user feedback events and learning corrections
| Gate | Original Decision | Feedback | Confidence | Timestamp |
|---|---|---|---|---|
No feedback events recordedFeedback will appear here when users interact with the Accountability Panel. |
||||
Gate Accuracy
Per-gate accuracy derived from user feedback corrections
How to Disagree with a Gate Verdict
To disagree with a gate verdict, use the Disagree button in the chat Accountability Panel. Each message's FORGEGATE results include an option to flag false positives, false negatives, or confirm correct decisions.
Your feedback is processed through the Shadow Teacher learning loop, which applies privacy-preserving differential noise before pattern detection. Systematic miscalibrations generate Calibration Genes automatically.
HIGH and CRITICAL stakes changes require manual approval in the Pending Approval Queue above. LOW and MEDIUM stakes genes are auto-approved by the Governance Engine.