Archie 17424808eb fix: require macOS 13.3+ with a clear reason ... llama.cpp release binaries link Accelerate's "new LAPACK"/ILP64 BLAS symbols (e.g. _cblas_sgemm$NEWLAPACK$ILP64), which Apple added in macOS 13.3. On older macOS (reported on an M1 running 12.6) llama-server aborts at dyld time with a cryptic "Symbol not found" — after the full ~3 GB setup has already run. ggml enables those interfaces unconditionally on Apple, so every pinned build has this floor and there is no older-macOS build to fall back to. Fail fast in assertSupportedPlatform() by checking the Darwin kernel version (22.4 ↔ macOS 13.3) with a plain-language message pointing at Software Update. Since the guard runs at the top of ensureBackend() on every boot, affected users see it before any download or sidecar start. Update README + submission docs to state the macOS 13.3 floor, and extend the platform-guard tests (old-kernel rejection, 13.3+/other-OS allowance). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>		2026-07-15 19:00:09 +02:00
..
superpowers	docs: note GPU auto-detection in README, changelog, and design spec	2026-07-09 19:38:26 +02:00
obsidian-submission.md	fix: require macOS 13.3+ with a clear reason	2026-07-15 19:00:09 +02:00
uru_screenshot_2.png	docs: replace README screenshot with chat-panel shot	2026-07-15 12:46:19 +02:00