v0.6.0
This release tightens the review guards so a reviewer can't sign off on their own work and can't mistake configuration for verified behaviour.
- Self-review is now blocked by authorship. The skill detects when the reviewer is the author of the work under review and prevents self-resolving findings on both review paths.
- Presence is not proof. Guidance now separates a config or setting being present from it being functionally verified — reviewers must confirm the behaviour, not just that the setting exists.
- New per-PR self-QA and read-findings guards to keep the review lifecycle honest.
- Added dogfood lessons captured from running a real review through the skill.
Full Changelog: v0.5.3...v0.6.0