See docs/DISCLAIMER_SNIPPET.md
Offline Installation Quickstart
This guide summarises how to install the project without internet access and run the Macro-Sentinel demo.
1. Build the wheelhouse
Run the helper script on a machine with connectivity:
./scripts/build_offline_wheels.sh
This collects wheels for all lock files inside a wheels/ directory. Set
SMOKE_ONLY=1 to build a minimal wheelhouse containing only the packages needed
for the smoke tests (numpy, PyYAML, pandas and prometheus_client).
2. Create lock files
Compile reproducible requirements from the wheel cache:
pip-compile --generate-hashes --output-file requirements.lock requirements.txt
pip-compile --no-index --find-links ./wheels --generate-hashes \
--output-file alpha_factory_v1/requirements-colab.lock \
alpha_factory_v1/requirements-colab.txt
3. Verify the environment
Use check_env.py to install any missing packages from the wheelhouse:
python check_env.py --auto-install --wheelhouse ./wheels
Environment variables
Set these variables when running offline so the helper scripts install packages from the local wheel cache and Insight never attempts network access:
| Variable | Purpose |
|---|---|
WHEELHOUSE |
Directory containing prebuilt wheels (e.g. ./wheels). |
AUTO_INSTALL_MISSING |
Set to 1 to automatically install missing packages. |
AGI_INSIGHT_OFFLINE |
Set to 1 to force local inference models. |
AGI_INSIGHT_BROADCAST |
Set to 0 to disable network broadcasting. |
4. Launch Macro-Sentinel
Start the demo with offline data feeds:
cd alpha_factory_v1/demos/macro_sentinel
LIVE_FEED=0 ./run_macro_demo.sh
See docs/OFFLINE_SETUP.md for additional details.