Request: Pre-built macOS Intel (x86_64) binary or universal build

[KrAsH-CoD3]

First off — JustHireMe is an incredible project. The local-first architecture, the graph visualization, and the semantic ranking are exactly what the ecosystem needs. Thank you for building this.

Problem

There's currently no pre-built installer for Intel Macs. The only macOS release asset is JustHireMe_1.0.0_aarch64.dmg (Apple Silicon). The JustHireMe.app.tar.gz is also compiled for aarch64, so macOS rejects it on Intel with "not supported on this Mac."
Building from source hits a hard blocker: lancedb>=0.30.2 has no macosx_*_x86_64 wheels — only arm64. uv sync fails with:
error: Distribution lancedb==0.30.2 can't be installed because it doesn't have a wheel for the current platform
hint: You're on macOS (macosx_15_0_x86_64), but lancedb only has wheels for ... macosx_11_0_arm64 ...

Impact

Intel Mac users can't run the app at all without:

1. Commenting out lancedb from pyproject.toml (loses semantic job matching)
2. Building the Tauri shell from source (requires Rust + full dev toolchain)

Request

Could the CI add an x86_64-darwin build target? Options:

• Use macos-13 (Intel) runner alongside macos-latest (ARM) in the release workflow
• Or build a universal binary with --target universal-apple-darwin
• Or make lancedb an optional dependency with a graceful fallback for macOS Intel (the code already has NullVectorStore — just needs the dependency marked optional) OR any better implementation.

Environment

• macOS 15.0 (Sequoia), Intel x86_64
• Python 3.13, Rust stable, uv latest.

Thank You!
Happy to help test or contribute a PR if you'd like.

[vasu-devs]

Your exact blocker is fixed on dev (commit 9548219): lancedb now carries an environment marker so uv sync skips the missing x86_64-macOS wheel, and the vector runtime already degrades gracefully without it (hashing-based matching; everything else works). Building from source on an Intel Mac works as of that commit.

Keeping this open for the second half of the ask - a pre-built Intel or universal binary. That needs an Intel CI runner (GitHub is retiring them) or cross-compilation of the PyInstaller sidecar, which is genuinely nontrivial; tracking it here.