Skip to main content

System Requirements & Platform Support

M31A is engineered for cross-platform portability across modern 64-bit operating systems.

Hardware Requirements​

ResourceMinimumRecommendedNotes
CPU2 Cores (x86_64 or ARM64)4+ CoresRequired for parallel DAG task execution and local compiler checks
RAM2 GB Available4 GB+ AvailableMemory footprint during SQLite indexing and AST analysis
Storage200 MB2 GB AvailableWorkspace databases, checkpoints, and artifact ledger staging
NetworkBroadband InternetLow-latency BroadbandOutbound HTTPS (TCP port 443) to NVIDIA NIM endpoint

Operating System Matrix​

M31A implementation platform coverage (verified in src/platform/ and .github/workflows/):

Operating SystemArchitectureSupport LevelSandbox MechanismStatus
Linux (Ubuntu 22.04+, Debian 12+, RHEL 9+, Arch)x86_64, aarch64Tier 1 (Primary)Landlock, seccomp, PID namespaces, /proc PID validationFull Production Support
macOS (12.0 Monterey, 13+, 14 Sonoma, 15 Sequoia)aarch64 (Apple Silicon), x86_64Tier 1sandbox-exec profile restrictions, posix_spawnFull Production Support
Windows (Windows 10, Windows 11, Server 2022)x86_64, arm64Tier 2Windows Job Objects, restricted token privilegesSupported (CLI/TUI)

Toolchain & Runtime Prerequisites​

When building from source:

  • Rust Toolchain: Rust 1.85.0 or later (Stable). Verified via rust-toolchain.toml.
  • Git: Version 2.30.0 or later. Required for worktree isolation, patch application, and trailer attribution.
  • SQLite: Bundled via SQLx with modern JSON and UUID v7 extensions. No external database server required.
  • Network Egress: Direct HTTPS connection to https://integrate.api.nvidia.com/v1. Proxies supported via HTTPS_PROXY.