Dockerfile

FROM python:3.10-slim

ENV PYTHONUNBUFFERED=1 \
    NODE_VERSION=20.x \
    # Set home to /app to avoid permission issues with /root in some environments
    HOME=/app

# 1. SYSTEM DEPS
RUN apt-get update && apt-get install -y \
    curl build-essential git git-lfs cmake ninja-build nginx netcat-openbsd \
    && curl -fsSL https://deb.nodesource.com/setup_$NODE_VERSION | bash - \
    && apt-get install -y nodejs \
    && rm -rf /var/lib/apt/lists/*

# 2. SETUP
WORKDIR /app
RUN git clone https://github.com/open-jarvis/OpenJarvis.git .

# 3. PYTHON SETUP
RUN pip install --no-cache-dir uv huggingface_hub
RUN uv sync --extra server
ENV CMAKE_ARGS="-DGGML_CPU=ON -DGGML_NATIVE=OFF -DGGML_AVX512=OFF -GNinja"
ENV CMAKE_BUILD_PARALLEL_LEVEL=8
RUN uv pip install --python /app/.venv llama-cpp-python[server]

# 4. FRONTEND BUILD
RUN cd frontend && \
    npm install --legacy-peer-deps && \
    npm install react-is --save-dev && \
    npx vite build --outDir dist --base ./

# 5. DOWNLOAD LOCAL MODEL
RUN mkdir -p /app/models && \
    python -c "from huggingface_hub import hf_hub_download; hf_hub_download(repo_id='bartowski/SmolLM2-1.7B-Instruct-GGUF', filename='SmolLM2-1.7B-Instruct-Q4_K_M.gguf', local_dir='/app/models', local_dir_use_symlinks=False)"

# 6. NGINX CONFIG
RUN echo 'pid /tmp/nginx.pid ;\nevents { worker_connections 768; }\nhttp {\n    include /etc/nginx/mime.types ;\n    server {\n        listen 7860;\n        location ~ ^/api/(.*)$ {\n            proxy_pass http://127.0.0.1:8000/$1$is_args$args;\n            proxy_set_header Host $host;\n        }\n        location / {\n            root /app/frontend/dist ;\n            try_files $uri $uri/ /index.html ;\n        }\n    }\n}' > /app/nginx.conf

# 7. STARTUP SCRIPT
RUN echo '#!/bin/bash\n\
# Configuration Setup\n\
MODEL_NAME=${OPENAI_API_MODEL:-"gpt-4o-mini"}\n\
CONF_DIR="/app/.openjarvis"\n\
mkdir -p "$CONF_DIR"\n\
\n\
echo "[model]\ndefault_model = \"$MODEL_NAME\"\n\n[[model.providers]]\nname = \"openai\"\napi_key = \"$OPENAI_API_KEY\"\n\n[[model.providers]]\nname = \"huggingface\"\ntoken = \"$HF_TOKEN\"\n\n[[model.providers]]\nname = \"llamacpp\"\nbase_url = \"http://127.0.0.1:8080/v1\"\nmodel_id = \"local-smollm2\"" > "$CONF_DIR/config.toml"\n\
\n\
# Start Llama.cpp (Local Inference)\n\
/app/.venv/bin/python -m llama_cpp.server --model /app/models/SmolLM2-1.7B-Instruct-Q4_K_M.gguf --host 127.0.0.1 --port 8080 > /app/engine.log 2>&1 &\n\
\n\
# Start Jarvis Brain\n\
/app/.venv/bin/jarvis serve --host 127.0.0.1 --port 8000 > /app/jarvis.log 2>&1 &\n\
\n\
# Health Check: Wait for Jarvis API to be ready\n\
echo "Waiting for Jarvis API..."\n\
MAX_RETRIES=30\n\
COUNT=0\n\
while ! curl -s http://127.0.0.1:8000/v1/models > /dev/null; do\n\
    sleep 2\n\
    COUNT=$((COUNT+1))\n\
    if [ $COUNT -ge $MAX_RETRIES ]; then echo "Jarvis failed to start"; exit 1; fi\n\
done\n\
\n\
echo "Backend ready. Starting Nginx..."\n\
exec nginx -c /app/nginx.conf -g "daemon off;"' > /app/start.sh && chmod +x /app/start.sh

EXPOSE 7860
CMD ["/app/start.sh"]