mirror of
https://github.com/proffesor-for-testing/agentic-qe.git
synced 2026-09-19 08:45:47 +08:00
c347662f63
Operationalize ADR-111's "competitive cheaper than frontier" / Ruv's ADR-179
quality-per-$ leaderboard as pure router primitives:
- value-score.ts: valueScore(m,{costWeight}) tunable slider, rankByValue, and
paretoFrontier (the non-dominated cost/quality set a router should choose from).
- MEASURED_QE_TEST_GEN: a dated D3/A12 economics snapshot (measured composite +
per-instance cost) — NOT vendor claims (qwen3-coder rank didn't transfer). The
measured frontier correctly drops qwen3:8b (below floor, dominated) and keeps
qwen3:30b-a3b (free, floor-clearing) + sonnet (frontier).
- 8 tests; strict-tsc clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>