Token efficiency Phase 3: description compression, adaptive max_tokens, conditional reflection#19

Merged

humancto merged 3 commits into

mainfrom

feat/token-efficiency-phase3

Mar 26, 2026

Commits on Mar 26, 2026

Token efficiency Phase 3: compress tool descriptions, adaptive max_tokens, conditional reflection

Archith
and
claude
committed
Fix review issues: restore behavioral hints, fix reflection metric, add tests
claude
committed
Fix P1: use routed provider for Ollama max_tokens override
claude
committed