Three-pass aggregate · 19 questions
- Mean judge score
- 3.93 → 4.04
- Keyword recall
- 0.579 → 0.580
- Mean request latency
- 14.7s → 23.3s
Baseline → planner · measured aggregate, limited scope
Planner and evaluation harness
TokenSmith Query Decomposition
- Public planner/evaluation snapshot
- Project status not reported
Across three passes on 19 questions, the planner reported a 4.04 vs 3.93 mean judge score, 0.580 vs 0.579 keyword recall, and 23.3s vs 14.7s mean request latency.