Rubric-to-Code Credit Assignment (RCCA) is a reinforcement-learning framework that converts functional feedback from rubrics into localized optimization signals for generated web-application code, using hierarchical rewards that distinguish format, source-code, runtime, and functional failures and align evaluator-generated attributions to specific code spans. The resulting Ling-RCCA-Flash model scores 41.25 on MiniAppBench and 76.19 on ArtifactsBench, surpassing prior baselines.
