Writing C++ with Jev by picking valid next tokens
The author experimented with TypeSafe's Jev decision/classifier model for C++: at each step it was given only grammatically valid next tokens to pick from, avoiding free-form generation.
Jev makes four decisions—context, routing, recovery, and testing—to cut coding agent costs by 68.8%, while Opus 5.5 writes the code.
4 decisions cut my coding agent bill by 68.8%. 1. Notes → what context does Opus actually need 2. Route → which tasks need Opus at all 3. Recover → retry, rethink or escalate 4. Tests → when to run them and which ones Jev decides all 4. Opus 5.5 writes the code. The