KiCad - the KiCad Bridge

@@ -1,6 +1,6 @@ {   "slug": "kicad-bridge",-  "version": "1.0.47",+  "version": "1.0.48",   "type": "app",   "description": "Skills for your container so your AI knows how to drive the KiCad bridge. The bridge runtime itself is the release zip; Adom Bridge loads that.",   "tags": [

@@ -4,7 +4,7 @@   "slug": "kicad-bridge",   "title": "KiCad - the KiCad Bridge",   "brief": "Reference implementation of the KiCad bridge — multi-instance Python server, forward path via kicad-cli, reverse path via in-process plugin. Most complex of the three bundled bridges.",-  "version": "1.0.47",+  "version": "1.0.48",   "tags": [     "kicad",     "pcb",

@@ -587,17 +587,32 @@ Implementation (win_taskbar.py):    window - the exact inverse of what it did before 0.9.230, both conclusions    derived from the same one-sentence meaning. -3. **Identity is not activity** (2026-08-23, caught on John's taskbar). The-   BASE icon (AUMID + iconPath, `_ensure_identity`) repairs the white-box shell-   cache we poisoned; it claims nothing about driving, so it must be-   UNCONDITIONAL: an always-on `_identity_sweeper` thread (20s) stamps every-   pcbnew/eeschema window, whoever opened it, verb or no verb. Before the-   sweeper, identity rode only on the badge path - so a window the user opened-   themselves (or one carrying the pre-.v2 stamp from an old bridge process)-   stayed a white box indefinitely, while a window a test verb had touched sat-   correct beside it. Badges and bars stay activity-gated; the base icon never-   is. Corollary: you cannot restamp identity by calling a passive verb --   that's rule 1 working as designed; the sweeper is the mechanism.+3. **This bridge stamps NO AUMIDs, ever** (John, 2026-08-23: "i just want+   default kicad behavior"). History: an early AI experiment stamped+   per-window AUMIDs on eeschema/pcbnew, at least once before icon extraction+   succeeded - Windows caches the icon PER AUMID STRING, so those buttons+   became permanent white boxes surviving KiCad reinstalls. Two generations of+   "fix" (fresh .v2 strings, then an always-on identity sweeper, 0.9.240-242)+   replaced the poison with Adom-flavoured identity - still an override, still+   wrong. The real fix (0.9.243) is REMOVAL: `_identity_repair()` runs once+   per boot, blanks the per-window AppUserModelID property (ctypes COM;+   VT_EMPTY - AD's set_window_identity has no clear path) on every+   eeschema/pcbnew window, deletes every HKCU `Adom.KiCad.*` registration,+   and hide/re-shows each button (never the foreground window) because Win11+   bakes tiles at button creation.+   The overlay badge and progress bar NEVER needed an AUMID - ITaskbarList3+   works per-hwnd on native buttons - so activity indication is unaffected.+   KNOW THE BASELINE before diagnosing: with all overrides removed, DEFAULT+   KiCad on Win11 with taskbar buttons combined shows a GENERIC WHITE icon for+   grouped standalone pcbnew/eeschema windows - a known upstream issue+   (kicad.info forum: "PCB Editor Icon turn white in taskbar if open as+   standalone app and using merge taskbar"). Verified on-box 2026-08-23: exe+   icon resource intact (ExtractAssociatedIcon renders), zero AUMIDs on any+   window, full icon-cache rebuild + explorer restart changed nothing, and no+   "Kicad.Kicad.*" string exists in any shipped binary - the installer stamps+   that AUMID onto the MAIN shortcut only, which is why kicad.exe's button+   renders and the editors' group doesn't. So a white editor-group icon is+   stock behaviour, not our damage; do not "fix" it without John asking.  ### The footprint stale-cache remedy is a TREE REFRESH, not a restart (2026-08-22, #51)