Artifact for "Beyond the Profiling Window: What Profile-Guided systemd Hardening Sees, and What It Breaks"

Scenario definitions, every generated policy, per-evaluation results, control runs, probe scripts with recorded outputs, the study harness, and the analysis code for an empirical study of profile-guided systemd hardening (SHH v2026.3.8, Ubuntu 26.04, systemd 259). Version 1.4 adds the study harness (code/harness/: guest provisioning, profiling, deployment checks, held-out runs, confirmation, the family-relaxation search, the post-hoc controls, and the export of the result and control tables, with its unit tests), the analysis code that regenerates every number of the paper from the released data (code/analysis/), and the helper behind three named scenario actions (scenarios/actions.py). It documents how scenario actions were run and how a policy is deployed (under a [Service] header), states that the reload scenarios check a value present before the reload, and adds the licence texts. Licences: the code (code/, scenarios/actions.py, and the shell scripts under mechanisms/) is MIT; the data, policies, recorded outputs, and documentation are CC BY 4.0; the unmodified systemd v259 files in code/harness/config/systemd/ are LGPL-2.1-or-later.