Governance Decisions in Open Collaboration Communities: Evidence Dataset

This dataset accompanies the manuscript "Accountability Governance in AI-Assisted Collaboration: Managerial Discretion and Intervention Boundaries in Open Collaboration Communities".It contains structured archival observations, source references, screening records, research classifications, alternative explanations, and verification materials supporting a qualitative comparative study of six open-source communities: Ghostty, Directus, pip, MDAnalysis, MolecularNodes, and Kornia. Polars is included as background material rather than as a seventh focal case.The package comprises nine CSV tables containing 140 records across different analytical units, documentation explaining the research methods and data fields, an evidence navigation guide, and a Python verification script.The dataset enables readers to inspect public source references, examine the screening and classification procedures, and check the principal evidence relationships discussed in the accompanying manuscript. It does not claim to reproduce the complete historical GitHub data collection process or independently replicate all qualitative judgments.License scope: CC BY 4.0 applies only to the original research annotations, tabular selections and English documentation that the depositing research authors have the right to license. The MIT License applies only to the original Python verification script, checks/verify_data.py. These licenses do not grant new rights in third-party GitHub code, discussion content, policy documents or linked websites. See LICENSE_AND_REUSE.md within the ZIP for the full scope and terms.