Linux Inspector: window-id capture and frame-geometry alignment#700
Linux Inspector: window-id capture and frame-geometry alignment#700mahbd wants to merge 2 commits into
Conversation
🔎 ZeuZ PR ReviewOpen the full report in ZeuZ: Review findings and apply suggestions
Agent breakdown→ General ReviewStatus: ✅ Completed The PR is directionally solid, but there is a notable correctness gap around → Security ReviewStatus: ✅ Completed No high-signal security issues found in the PR diff. The changes add Linux window enumeration and screenshot alignment logic, plus pinned dependencies, without introducing a clear injection, authz, or secret-handling flaw. → Performance ReviewStatus: ✅ Completed I found one performance risk: the new Linux apps/window discovery path now shells out to → Testing ReviewStatus: ✅ Completed The Linux inspector change adds new window-aware behavior and a new
|
🔎 ZeuZ PR ReviewOpen the full report in ZeuZ: Review findings and apply suggestions
Agent breakdown→ General ReviewStatus: ✅ Completed I found one correctness issue in the new window-selection fallback, plus one maintainability issue in the API model definition. → Security ReviewStatus: ✅ Completed No high-signal security issues found in this PR diff. The changes add window selection and geometry alignment without introducing obvious injection, authz, or secret-handling regressions. → Performance ReviewStatus: ✅ Completed The PR adds useful multi-window support, but it also introduces several extra xdotool round-trips on the hot paths for app lookup and → Testing ReviewStatus: ✅ Completed The PR adds important Linux window-selection and geometry-alignment logic, but there are no tests covering the new code paths in
|
…geometry alignment Crops composite captures to the AT-SPI frame bounds so screenshots align with accessibility-tree coordinates, and allows callers to target a specific window by id instead of just app name.
1d36d8e to
a3c40a7
Compare
🔎 ZeuZ PR ReviewOpen the full report in ZeuZ: Review findings and apply suggestions
Agent breakdown→ General ReviewStatus: ✅ Completed I found one high-impact issue in the new screenshot alignment logic: the crop rectangle is computed from AT-SPI desktop coordinates but compared against the captured window's local geometry, so the alignment path will often never apply. I also noted a smaller maintainability issue around a mutable model default. → Security ReviewStatus: ✅ Completed No security-relevant issues found in the PR diff; the changes are focused on window selection and screenshot alignment without introducing obvious injection, auth, or data-exposure risks. → Performance ReviewStatus: ✅ Completed One performance concern: → Testing ReviewStatus: ✅ Completed The PR adds new window-selection and frame-alignment behavior, but I don’t see any tests covering the new Linux inspector paths or the regression scenario they fix.
|
Summary
window_idparameter tocapture_screenshotandget_ui_tree, allowing a specific window to be targeted directly instead of relying on app-name lookup._get_frame_geometry_for_windowto find the AT-SPI frame matching a window and crop composite screenshot captures to its bounds, fixing a ~40px misalignment between bounding boxes and screenshot content (window decoration/title bar offset).window_idare unaffected — they fall back to the previous app-name-based behavior.Test plan