llama.cpp b11199 (86a24a182bd3c6c877aaf3734d99f9a1820f8cbd) for x86_64, built in Ubuntu 22.04.5 LTS (glibc 2.35) for PoolParty by
scripts/build-llama-cpp.sh's own flags less the macOS-only -DGGML_METAL=ON: static (BUILD_SHARED_LIBS=OFF), GGML_RPC=ON,
GGML_NATIVE=OFF, no tests, no examples, no web UI (LLAMA_BUILD_UI=OFF, LLAMA_USE_PREBUILT_UI=OFF). The rpc target is
ggml-rpc-server and installs as rpc-server, the name the agent expects. CPU backend only: no GPU backend is enabled and
none is claimed. THE GLIBC FLOOR IS THIS BASE'S.
