The patent system rests on a fundamental bargain: temporary monopoly rights in exchange for the disclosure of technical knowledge. Yet, the complexity of transferring technical know-how often limits the effectiveness of disclosure, keeping innovation highly localized. We study whether actionable disclosure in the form of publicly available executable code can overcome these frictions. We rely on a novel dataset linking two decades of patenting activity in the United States to contributions in public code repositories by 1,556 organizations. Using neural language models, we measure the semantic similarity between patent descriptions and public code contributions by these firms to identify patents with high digital disclosure. We find that these patents attract citations from inventors located approximately 17\% farther away than those citing a group of control patents, suggesting that actionable disclosure in the form of executable code reduces spatial barriers to knowledge diffusion.