Replication package for: Who Checks the Agents? Verification Topology and the Disclosure Gap in Documented LLM Agent Populations

Protocol files with addenda A1 to A4 and A6, codebooks (v2.2, v2.3, coding sheet 2.4), coded case data with verbatim excerpts, adjudications, the frozen data sets v1.1 (15 cases, pre-specified test) and v2 (42 cases, post hoc extension E1), codings of four language-model coders from three vendors, excerpt-relevance judgements, csQCA analysis code and results, descriptive analyses of E1, the post hoc analyses of addendum A3, the corrections and the blind recoding of the checkpoint fields of 54 cases with three adjudications (addendum A6), the external audit of the coder divergence, three later AI reader tests on 44 cases (A7 to A9), the human anchor sample, Supplementary Material S1 (full appendices of the manuscript before shortening), a post hoc decomposition of missing checkpoint values by outcome, a rerun of the configurational analysis with the first author's anchor values, the third post-window news check of 27 October 2026 with its cross-check, and R implementation comparisons of v1.1 and E1, for the article of the same title (prepared for submission to the Journal of Artificial Intelligence Research, Special Track AI and Society). Coding was carried out by language models; see README.md. Source full texts are not redistributed; intermediate run files are regenerated by the documented commands.