AI Teammates as Authors and as Reviewers: replication package (event logs, analysis, classifier audit)

Replication package for AI Teammates as Authors and as Reviewers: Human Code Review Across Projects and Over Time (submitted to Empirical Software Engineering, special issue on agentic software engineering, 2026). Contents: pseudonymised pull-request event logs of seventeen open-source projects on GitHub and GitLab (2024–2026), the analysis drivers and the result tables every number in the article is generated from, the audit of the AI-involvement classifier (protocol, prompt, blind sheets, three models' labels, scores), the article and its supplement with their generators, tables and figures, the collection configurations and the source of the gitlab-activity package. Human logins are replaced by pseudonyms; no pull-request body, commit message, comment or diff content was collected. See README.md in the archive.