# ══════════════════════════════════════════════════════════════════════════════
# Zebrunner MCP Server - Multi-stage Docker Build
# ══════════════════════════════════════════════════════════════════════════════
# Supports: stdio transport (default), video/image analysis (ffmpeg, sharp, tesseract)
# Usage: docker build -t msarychau/mcp-zebrunner:latest .
# ══════════════════════════════════════════════════════════════════════════════
ARG NODE_VERSION=20
# ──────────────────────────────────────────────────────────────────────────────
# Stage 1: Builder - Compile TypeScript and install dependencies
# ──────────────────────────────────────────────────────────────────────────────
FROM node:${NODE_VERSION}-alpine AS builder
# Install build dependencies for native modules (sharp, etc.)
RUN apk add --no-cache \
python3 \
make \
g++ \
vips-dev
WORKDIR /build
# Copy package files first (leverage Docker layer caching)
COPY package.json package-lock.json ./
# Install ALL dependencies (including devDependencies for TypeScript compilation)
RUN npm ci
# Copy source code and config files
COPY tsconfig.json ./
COPY src/ ./src/
# Build TypeScript to JavaScript
RUN npm run build
# Prune devDependencies for smaller production image
RUN npm prune --production
# ──────────────────────────────────────────────────────────────────────────────
# Stage 2: Production - Minimal runtime image
# ──────────────────────────────────────────────────────────────────────────────
FROM node:${NODE_VERSION}-alpine AS production
# Labels for Docker Hub and container identification
LABEL org.opencontainers.image.title="Zebrunner MCP Server"
LABEL org.opencontainers.image.description="MCP server for Zebrunner TCM - test cases, suites, coverage analysis, launchers, and reporting"
LABEL org.opencontainers.image.vendor="Maksim Sarychau"
LABEL org.opencontainers.image.url="https://github.com/maksimsarychau/mcp-zebrunner"
LABEL org.opencontainers.image.source="https://github.com/maksimsarychau/mcp-zebrunner"
LABEL org.opencontainers.image.licenses="AGPL-3.0"
LABEL ai.mcp.server="true"
LABEL ai.mcp.transport="stdio"
# Install runtime dependencies for native modules
# - vips: required by sharp for image processing
# - ffmpeg: required for video analysis
# - tesseract-ocr: required for OCR in screenshots
# - ca-certificates: for HTTPS requests
# - wget: for downloading tessdata
RUN apk add --no-cache \
vips \
ffmpeg \
tesseract-ocr \
ca-certificates \
wget \
&& rm -rf /var/cache/apk/*
# Download tesseract training data from official repository
RUN mkdir -p /usr/share/tessdata && \
wget -q -O /usr/share/tessdata/eng.traineddata \
https://github.com/tesseract-ocr/tessdata_best/raw/main/eng.traineddata
# Create non-root user for security
RUN addgroup -g 1001 -S mcpuser && \
adduser -u 1001 -S mcpuser -G mcpuser
WORKDIR /app
# Copy built artifacts from builder stage
COPY --from=builder --chown=mcpuser:mcpuser /build/dist ./dist
COPY --from=builder --chown=mcpuser:mcpuser /build/node_modules ./node_modules
COPY --from=builder --chown=mcpuser:mcpuser /build/package.json ./
# Copy default rules file if it exists (use wildcard to avoid build failure)
COPY --chown=mcpuser:mcpuser mcp-zebrunner-rules.md* ./
# Create directory for optional config mounts
RUN mkdir -p /config && chown mcpuser:mcpuser /config
# Switch to non-root user
USER mcpuser
# Environment variables with defaults
ENV NODE_ENV=production
ENV DEBUG=false
ENV EXPERIMENTAL_FEATURES=false
ENV MAX_PAGE_SIZE=100
ENV DEFAULT_PAGE_SIZE=10
ENV ENABLE_RULES_ENGINE=false
# Required environment variables (must be provided at runtime)
# ENV ZEBRUNNER_URL=
# ENV ZEBRUNNER_LOGIN=
# ENV ZEBRUNNER_TOKEN=
# Start MCP server (stdio transport)
# Note: Use 'docker run --init' for proper signal handling instead of tini
CMD ["node", "dist/server.js"]