Dynamic escalation on low scores

{
  "code": "Dynamic Escalation on Low Scores — IMPLEMENTED.\n\nFile modified: c8_platform/foundry_pipeline.py (3 edits)\n\nChanges:\n1. Added `effort,quality` to failed task query SELECT (line 526) — needed to read current effort tier and quality score\n2. Added effort escalation logic before retry payload (lines 546-560):\n   - Defines effort ladder: low → medium → high → max\n   - If quality.overall < 0.50, bumps to next tier\n   - Records `escalated_from` and `escalation_reason` in retry metadata\n   - At max tier, stays at max (no further escalation)\n   - If score >= 0.50 or not in ladder, keeps current effort\n3. Added `effort: escalated_effort` to retry payload (line 580)\n\nEscalation tracking in metadata:\n- `escalated_from`: original effort tier (e.g., \"low\")\n- `escalation_reason`: e.g., \"score_0.45_below_0.50\"\n- `retry_count`: incremented as before\n\nCost delta tracking: The effort tier already maps to cost_class in EFFORT_TIERS (foundry_bridge.py lines 141-162):\n- low: ~$0.001/task\n- medium: ~$0.01/task\n- high: ~$0.10/task\n- max: ~$0.15/task\n\nTests: 21/21 passed (test_foundry_retrospective.py), no regressions.",
  "content": "Dynamic Escalation on Low Scores — IMPLEMENTED.\n\nFile modified: c8_platform/foundry_pipeline.py (3 edits)\n\nChanges:\n1. Added `effort,quality` to failed task query SELECT (line 526) — needed to read current effort tier and quality score\n2. Added effort escalation logic before retry payload (lines 546-560):\n   - Defines effort ladder: low → medium → high → max\n   - If quality.overall < 0.50, bumps to next tier\n   - Records `escalated_from` and `escalation_reason` in retry metadata\n   - At max tier, stays at max (no further escalation)\n   - If score >= 0.50 or not in ladder, keeps current effort\n3. Added `effort: escalated_effort` to retry payload (line 580)\n\nEscalation tracking in metadata:\n- `escalated_from`: original effort tier (e.g., \"low\")\n- `escalation_reason`: e.g., \"score_0.45_below_0.50\"\n- `retry_count`: incremented as before\n\nCost delta tracking: The effort tier already maps to cost_class in EFFORT_TIERS (foundry_bridge.py lines 141-162):\n- low: ~$0.001/task\n- medium: ~$0.01/task\n- high: ~$0.10/task\n- max: ~$0.15/task\n\nTests: 21/21 passed (test_foundry_retrospective.py), no regressions."
}