Обнаружение доступных системных ресурсов

★ 7.0 · productivity

get-available-resources is a Claude Code skill that builds a conservative picture of system resources available to the current process before resource-sensitive workloads. It separately tracks host CPU topology, process affinity, cgroup v2 limits, Slurm scheduler allocation, and accelerator candidates — NVIDIA CUDA, AMD ROCm, and Apple Metal — then emits a redacted JSON snapshot without running stress tests, benchmarks, or writing files by default. The bundle includes `detect_resources.py` and `plan_workload.py` scripts, an optional psutil 7.2.2 dependency for broader cross-platform coverage, a `--skip-accelerators` flag to reduce probe latency, and strict safety rules that redact hostnames, absolute paths, device UUIDs, and PCI addresses. Useful for developers and data engineers who need to plan parallelism, size CPU worker pools, or check GPU availability inside containers and HPC scheduler allocations. Requires Python 3.11+ on Linux, macOS, or Windows.