ContractRL: shielded contract-constrained RL for auditable JSON tool-call repair
This arXiv paper (arXiv:2610.00328v1) introduces ContractRL, a contract-constrained sequential repair protocol that models verifier-guided JSON repair as a bounded decision process with action masks derived from schemas (RFC-6902) and an immutable repair history for auditability. In evaluations over multiple seeds, ContractRL achieved 0.9362 semantic success with 34.4 generated tokens compared with 0.9076 and 44.9 tokens for Patch-SFT and 0.9148 and 137.2 tokens for full regeneration; policy optimization further raised semantic success from 0.9186 (supervised ContractRL) to 0.9375, and analyses link gains to localized corrections while adversarial and multi-turn tests expose remaining failure modes.