| name | setup-trtmc-environment |
| description | Prepare a TensorRT-Model-Connect development or deployment-validation environment from a fresh checkout on an unfamiliar host. Use before builds, tests, packaging, or runtime work when no working repo environment is known. |
Set Up The Environment
Start from the checkout, not from existing container names or machine-specific
workspace conventions.
- Find the repository root with
git rev-parse --show-toplevel.
- Read the current environment guide,
Dockerfile, and matching
scripts/docker_build_*.sh / scripts/docker_run_*.sh. Treat them as the
source of truth; do not copy assumptions from another host.
- Inspect the host as needed (
uname, docker info, nvidia-smi, free disk)
and select a repo-supported path. If none matches, explain the gap and stop.
Do not install drivers or reconfigure the container runtime without separate
authorization.
- Build the selected repo image when it is missing or stale, using the repo's
build path. Do not prebuild unrelated model-family reference profiles.
- Start a fresh container from that image with this checkout mounted
read-write for development or read-only for deployment validation. Choose
paths and names for this checkout; do not reuse, migrate, or remove unrelated
containers.
- Verify the checkout mount, required GPU visibility, and TensorRT import, then
run the requested build, test, packaging, or runtime command.
Keep model-specific dependencies on demand and follow each family's own lock
and verification files. Report setup evidence separately from compilation,
tests, model parity, performance, and production-deployment evidence.
