Shadow Daily Reflection
v1.1Shadow — Daily Reflection · owner shadow-team · updated 2026-05-04
Weight sum
1.00
Dimensions
10
Safety gates
2
Runs using this
2
Safety gates:pii leakagemedical advice without disclaimer
Dimension analysis
4 cases · 2 runs| Dimension | Method | Weight | Threshold | Avg score | Pass rate | Performance |
|---|---|---|---|---|---|---|
Life-area classification accuracy life_area_accuracy | Deterministic | 0.15 | ≥0.80 | 0.93 | 100%(4) | |
Emotional nuance emotional_nuance | LLM Judge | 0.10 | ≥0.70 | 0.87 | 100%(4) | |
Non-judgmental tone non_judgmental_tone | LLM Judge | 0.10 | ≥0.75 | 0.93 | 100%(4) | |
Useful next step useful_next_step | LLM Judge | 0.10 | ≥0.65 | 0.77 | 75%(4) | |
Memory relevance memory_relevance | Claim Pipeline | 0.10 | ≥0.70 | 0.80 | 75%(4) | |
Completeness completeness | LLM Judge | 0.10 | ≥0.70 | 0.85 | 100%(4) | |
Hallucination risk hallucination_risk | Claim Pipeline | 0.15 | ≥0.80 | 0.89 | 100%(4) | |
Tone fit tone_fit | LLM Judge | 0.05 | ≥0.70 | 0.91 | 100%(4) | |
Consistency consistency | LLM Judge | 0.05 | ≥0.70 | 0.94 | 100%(4) | |
Actionability actionability | LLM Judge | 0.10 | ≥0.65 | 0.74 | 75%(4) |
Live LLM Scorer
Helpfulness · GPT-4o-mini · real API call
2000 chars left