Dockerfile

# SimReady Validator — HuggingFace Space image
#
# Phase-3 of the PRD: validation runs where the dataset lives. The Space
# downloads a target dataset via huggingface_hub, runs the bundled
# simready-report validator with `--no-use-kit` (no Isaac Sim on HF
# hardware), and writes the verdict back to the dataset as a PR.
#
# Image strategy: pre-bake all heavy deps (usd-core, omni.asset_validator,
# simready-validate) and the simready-foundation specs checkout into the
# image so cold-start is fast — the per-validation cost is just the
# dataset download + USD parsing, not pip resolution.

FROM python:3.11-slim

# Base build deps. usd-core ships wheels for x86_64 so we don't need
# build-essential, but git is required to clone simready-foundation at
# build time and rsync is handy for the dashboard `_static/` mirror.
# Node 20 is for the Claude Code CLI (@anthropic-ai/claude-code) that
# the recovery layer in runner.py invokes when the validator crashes.
RUN apt-get update && apt-get install -y --no-install-recommends \
        git curl ca-certificates rsync gnupg \
    && curl -fsSL https://deb.nodesource.com/setup_20.x | bash - \
    && apt-get install -y --no-install-recommends nodejs \
    && rm -rf /var/lib/apt/lists/*

# Claude Code CLI. Pinned to a known-good version; bump deliberately
# alongside the recovery skill's wording if the SDK changes shape.
RUN npm install -g @anthropic-ai/claude-code@latest

# HF Spaces run as a non-root user (UID 1000). Match it so the volume
# mounts are writable without permission gymnastics.
ARG USER=appuser
RUN useradd -m -u 1000 ${USER}

WORKDIR /home/${USER}/app

# Python deps first so they layer-cache independently of the validator
# code. Pinned to the same versions install-simready-sdk.sh uses on the
# DGXC runner (single source of truth for "what runs the validator").
COPY tools/hf_space/requirements.txt ./requirements.txt
RUN pip install --no-cache-dir -r requirements.txt

# Foundation specs: pinned to a specific commit so spec churn upstream
# doesn't break the Space's compatibility with the pinned simready-
# validate package. The validator's loader rejects features whose
# requirement codes aren't registered, and the foundation repo's main
# branch has moved ahead of what simready-validate 2026.4.x understands
# — sticking to the 2026.04.0 release tag (commit 805d2c5) matches the
# DGXC runner's working setup. Bump deliberately when both the spec
# and the validator package are tested together.
ENV SIMREADY_FOUNDATIONS_PATH=/opt/simready_foundations
ENV SIMREADY_FOUNDATIONS_COMMIT=805d2c50179a9878c89b0f41baaa0ecafe47c3d7
RUN git clone https://github.com/NVIDIA/simready-foundation \
        ${SIMREADY_FOUNDATIONS_PATH} \
    && cd ${SIMREADY_FOUNDATIONS_PATH} \
    && git checkout ${SIMREADY_FOUNDATIONS_COMMIT} \
    && chown -R ${USER}:${USER} ${SIMREADY_FOUNDATIONS_PATH}

# Vendored simready-foundation-core wheel. Built offline from the
# foundation repo at commit 805d2c5 using:
#   1. python -m omni.usd_profiles.codegen --docs-root docs
#        --destination-dir _build/python --namespace omni.capabilities
#   2. python -m build --wheel nv_core/sr_specs
# Plus a local patch (docs/__init__.py + pyproject.toml entry-point
# declaration) so the wheel ships the SimReadyPlugin entry-point that
# omni.asset_validator's allow-list discovery requires. Without this
# wheel the validator loads only DefaultPlugin and ends up with
# profiles=0. The clone above is still needed for spec markdown
# (SIMREADY_FOUNDATIONS_PATH) — wheel provides the runtime package.
COPY tools/hf_space/wheels/simready_foundation_core-0.2.0a1-py3-none-any.whl /tmp/
RUN pip install --no-cache-dir /tmp/simready_foundation_core-0.2.0a1-py3-none-any.whl \
    && rm /tmp/simready_foundation_core-0.2.0a1-py3-none-any.whl

# Copy the bundled validator (the same code our DGXC runner uses) and
# the Space's own entry points. Keep the in-repo path so the validator's
# relative imports (report.py, external_deps.py) still resolve.
COPY tools/validation /home/${USER}/app/tools/validation
COPY tools/hf_space/app.py     ./app.py
COPY tools/hf_space/runner.py  ./runner.py
COPY tools/hf_space/github_issues.py ./github_issues.py

# Claude Code skills, mounted where the CLI auto-discovers them. The
# simready-validator-agent skill drives recovery + issue-filing; the
# simready-report skill is the validator wrapper Claude already knows
# from the local /simready-report workflow.
RUN mkdir -p /home/${USER}/.claude/skills
COPY tools/hf_space/skills/simready-validator-agent /home/${USER}/.claude/skills/simready-validator-agent
COPY tools/validation/plugins/simready-report/skills/simready-report /home/${USER}/.claude/skills/simready-report

RUN chown -R ${USER}:${USER} /home/${USER}/app /home/${USER}/.claude

USER ${USER}

# HF Spaces with the Docker SDK serve whatever listens on $PORT
# (default 7860). Gradio honors GRADIO_SERVER_PORT.
ENV GRADIO_SERVER_NAME=0.0.0.0
ENV GRADIO_SERVER_PORT=7860
EXPOSE 7860

CMD ["python", "app.py"]