ODS V3 Pre-Release: Public testing and refinement ahead of the official V3 launch. Turn your PC, Mac, or Linux box into a private AI server.
Python ai-agentsamdcomfyuidockerllama-cppllmlocal-ain8nnvidiaopen-webuiragself-hostedspeech-to-textstrix-halotext-to-speechworkflow-automation
#!/usr/bin/env bash
# Run the real requirements phase and engine against fixture files only.
# The complete GPU evidence of a host-native llama-server (the Windows
# Portal's llama-server.exe) replaces only the CPU warning.
# Variables below are consumed by the sourced requirements phase.
# shellcheck disable=SC2034
set -euo pipefail
ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
tmp="$(mktemp -d)"
trap 'rm -rf "$tmp"' EXIT
mkdir -p "$tmp/source/scripts" "$tmp/source/lib"
cp "$ROOT/scripts/preflight-engine.sh" "$tmp/source/scripts/preflight-engine.sh"
cp "$ROOT/lib/safe-env.sh" "$tmp/source/lib/safe-env.sh"
chmod +x "$tmp/source/scripts/preflight-engine.sh"
touch "$tmp/source/docker-compose.base.yml"
run_case() (
local label="$1" url="$2" name="$3" vram="$4" gpu_status="$5"
local backend="${6:-cpu}" ram="${7:-64}" disk="${8:-200}"
SCRIPT_DIR="$tmp/source" INSTALL_DIR="$tmp/install"
LOG_FILE="$tmp/requirements.log" PREFLIGHT_REPORT_FILE="$tmp/preflight.json"
TIER=2 RAM_GB="$ram" DISK_AVAIL="$disk"
GPU_BACKEND="$backend" GPU_VRAM=0 GPU_NAME=None GPU_COUNT=0
INTERACTIVE=false DRY_RUN=true PYTHON_CMD=python3
CAP_PLATFORM_ID=wsl CAP_COMPOSE_OVERLAYS=docker-compose.base.yml
ENABLE_VOICE=false ENABLE_WORKFLOWS=false ENABLE_RAG=false ENABLE_COMFYUI=false
EXTERNAL_LLM_URL='' ODS_MODE=local
# Deliberately not exported: phase 04 must pass this structured evidence.
NATIVE_LLM_BASE_URL="$url" NATIVE_LLM_GPU_NAME="$name" NATIVE_LLM_GPU_VRAM_MB="$vram"
export -n NATIVE_LLM_BASE_URL NATIVE_LLM_GPU_NAME NATIVE_LLM_GPU_VRAM_MB
declare -A SERVICE_PORTS=()
tier_rank() { printf '2\n'; }
ods_progress() { :; }; chapter() { :; }; log() { :; }
ai_ok() { printf 'OK: %s\n' "$*"; }; ai_bad() { printf 'BAD: %s\n' "$*"; }
ai_warn() { printf 'WARN: %s\n' "$*"; }; warn() { printf 'WARN: %s\n' "$*"; }
# Prevent port/process inspection from touching the running installation.
docker() { :; }; pgrep() { return 1; }; lsof() { return 1; }
ss() { :; }; netstat() { :; }
source "$ROOT/installers/phases/04-requirements.sh" >"$tmp/output"
python3 - "$PREFLIGHT_REPORT_FILE" "$tmp/output" "$gpu_status" "$backend" \
"$ram" "$disk" "$PREFLIGHT_WARNINGS" "$REQUIREMENTS_MET" <<'PY'
import json
import pathlib
import sys
report_path, output_path, status, backend, ram, disk, warnings, requirements_met = sys.argv[1:]
report = json.loads(pathlib.Path(report_path).read_text())
output = pathlib.Path(output_path).read_text()
checks = {check["id"]: check for check in report["checks"]}
gpu = checks["gpu-vram" if backend == "nvidia" else "gpu-backend"]
assert gpu["status"] == status, gpu
assert report["inputs"]["gpu_backend"] == backend, "local backend must remain unchanged"
if status == "pass":
assert "Host-native llama-server GPU route configured" in gpu["message"], gpu
assert report["inputs"]["external_gpu"] == {
"provider": "host-native-llama-server", "gpu_name": "AMD Radeon RX 9070 XT", "gpu_vram_mb": 16304,
}
assert "CPU fallback selected" not in output, output
elif backend == "cpu":
assert report["inputs"]["external_gpu"] is None
assert "CPU fallback selected" in output, output
else:
assert "no NVIDIA GPU VRAM was detected" in output, output
expected_warnings = int(status == "warn") + int(int(ram) < 32)
expected_blockers = int(int(disk) < 50)
assert report["summary"]["warnings"] == expected_warnings, report["summary"]
assert int(warnings) == expected_warnings, warnings
assert report["summary"]["blockers"] == expected_blockers, report["summary"]
assert requirements_met == str(expected_blockers == 0).lower(), requirements_met
if int(ram) < 32:
assert checks["memory"]["status"] == "warn" and "RAM 16GB" in output, output
if int(disk) < 50:
assert checks["disk"]["status"] == "blocker" and "Disk 10GB" in output, output
PY
printf 'PASS: %s\n' "$label"
)
gpu_name='AMD Radeon RX 9070 XT'
endpoint='http://localhost:8080'
run_case 'Host-native AMD route replaces the CPU fallback warning' "$endpoint" "$gpu_name" 16304 pass
run_case 'Host-native route keeps RAM warnings and disk blockers' "$endpoint" "$gpu_name" 16304 pass cpu 16 10
run_case 'Missing native endpoint retains CPU fallback warning' '' "$gpu_name" 16304 warn
run_case 'Missing GPU name retains CPU fallback warning' "$endpoint" '' 16304 warn
run_case 'Unknown GPU name retains CPU fallback warning' "$endpoint" Unknown 16304 warn
run_case 'Zero VRAM retains CPU fallback warning' "$endpoint" "$gpu_name" 0 warn
run_case 'Invalid VRAM retains CPU fallback warning' "$endpoint" "$gpu_name" 16GB warn
run_case 'Negative VRAM retains CPU fallback warning' "$endpoint" "$gpu_name" -1 warn
run_case 'Other GPU warnings remain visible with a host-native route' "$endpoint" "$gpu_name" 16304 warn nvidia