Pace–procedure model of AI-ready institutional capacity: coding data, corpus and code for a comparative study of Ukraine and EU candidate countries (2019–2026)

Replication package for a manuscript under double-anonymised peer review on artificial intelligence in Ukraine's public administration, compared with four EU candidate countries (Serbia, Moldova, Albania and North Macedonia). The study proposes a pace–procedure model of AI-ready institutional capacity. A country is described by two measurable coordinates: the pace of digital service delivery, computed as the mean global percentile of three international indicators (UN Online Service Index, UN E-Participation Index, Oxford Insights Public Sector Adoption pillar), and the procedural depth of AI governance, computed from a documentary coding matrix of twelve codes derived from the EU AI Act, the Council of Europe Framework Convention on AI (CETS 225), the HUDERIA methodology and the OECD framework for trustworthy AI in government, in an enacted and a drafted layer. Documents were fixed as of 30 September 2026. The record contains: the coding book (21 codes in six groups, with sources and scoring rules); the document corpus (35 legal, strategic and policy documents of the five countries, with layer and codes informed); two independent codings and the reconciled coding matrix (16 codes × 5 countries × 2 layers); article-level coding of Albania's draft Law "On Artificial Intelligence" (public consultation, 29 May – 26 June 2026); global percentile indicators of digital service delivery and the derived model values (pace, procedural depth, declarative ratio, gap); Python scripts that reproduce the model indicators, inter-coder reliability statistics (weighted Cohen's κ, Krippendorff's α), Monte Carlo weight sensitivity (10,000 Dirichlet draws), leave-one-code-out checks and Figures 4–6; VOSviewer inputs for the bibliometric maps: two Scopus corpora (3,057 and 485 records, bibliographic metadata only), the thesaurus file (177 rules) and the JSON export of the focused-corpus map; the figures as submitted. Third-party datasets used for the percentile computation (UN E-Government Survey 2024, Oxford Insights Government AI Readiness Index 2025, World Bank GovTech Maturity Index 2025) are not redistributed; the README explains where to obtain them and how to place them for full reproduction. Derived values for the five countries are included so that the model scripts run without the raw files. Data and documentation are licensed under CC BY 4.0; code under the MIT licence. This is an anonymous deposit for the period of double-anonymised review; author names, affiliations and the link to the published article will be added in a new version of the record upon acceptance.