This replication package accompanies the manuscript titled “Agentic Capital as a Productive Asset in the Agentic Economy: A Panel Econometric Analysis of Productivity Dynamics.” The package provides the data structure, variable dictionary, reproducibility files, scripts, extracted manuscript tables, econometric outputs, diagnostics, and robustness materials used to support the empirical architecture of the article. The study develops Agentic Capital as a measurable productive category composed of six pillars: Model Capital, Data Capital, Robotic Physical Capital, Compute Capital, Protocol Capital, and Human Sovereignty Capital. It applies a country-year panel framework for nine economies over 2005–2024 and examines the relationship between Agentic Capital and labour productivity growth. The replication materials include raw-data templates, cleaned-panel templates, pillar-construction logic, index-construction formulas, econometric specification files, diagnostic outputs, robustness-check files, table inventories, and documentation. The package is designed to make the construction of the Agentic Capital indicators transparent, auditable, and reproducible, while also clarifying the first-generation and proxy-based nature of the empirical implementation. The files support the article’s methodological framework, including additive and geometric Agentic Capital indices, the Gondauri-adjusted Agentic Capital Index, the Balanced Agentic Capital Ratio, the Agentic Capital Gap, alternative resilience-anchor robustness checks, fixed-effects specifications, Driscoll-Kraay robust inference, IV-2SLS, dynamic GMM, quantile regression, threshold analysis, mediation analysis and component-level fixed-effects diagnostics. This repository is intended to improve transparency, reproducibility, and future extensibility. Future researchers may use the package to replicate the reported structure, update the dataset with additional countries or more direct pillar-level indicators, and extend the Agentic Capital framework to broader country, sectoral, or firm-level applications.