# Dockerfile.standard - MCP Server with Semantic Search
# Features: All minimal features + AI-powered semantic search, GitHub artifact sync
FROM mcp-index:minimal as base
# Switch to root for installation
USER root
# Install additional Python dependencies for semantic search
RUN pip install --no-cache-dir \
# Semantic search dependencies
voyageai>=0.2.0 \
qdrant-client>=1.7.0 \
# Reranking dependencies
sentence-transformers>=2.0.0 \
# Enhanced processing
numpy>=1.24.0 \
# Artifact management
requests>=2.31.0 \
PyGithub>=2.1.0
# Final stage
FROM base
# Copy configuration templates
COPY --chown=mcp:mcp docker/config/standard.env /app/config/
COPY --chown=mcp:mcp docker/scripts/entrypoint-standard.sh /app/
# Set environment variables for standard configuration
ENV PYTHONPATH=/app \
MCP_WORKSPACE_ROOT=/workspace \
DATABASE_URL=sqlite:////app/data/code_index.db \
# Enable semantic search (requires API key)
SEMANTIC_SEARCH_ENABLED=true \
SEMANTIC_EMBEDDING_MODEL=voyage-code-3 \
SEMANTIC_COLLECTION_NAME=code-embeddings \
SEMANTIC_VECTOR_SIZE=1024 \
SEMANTIC_SEARCH_LIMIT=20 \
# Reranking configuration
RERANKING_ENABLED=true \
RERANKER_TYPE=cross-encoder \
# GitHub artifact sync enabled by default (configurable)
MCP_ARTIFACT_SYNC=true \
AUTO_UPLOAD=true \
AUTO_DOWNLOAD=true \
ARTIFACT_RETENTION_DAYS=30 \
# Performance settings
INDEXING_MAX_WORKERS=8 \
INDEXING_BATCH_SIZE=50 \
INDEXING_GENERATE_EMBEDDINGS=true \
# Cache settings
CACHE_BACKEND=memory \
QUERY_CACHE_ENABLED=true \
QUERY_CACHE_DEFAULT_TTL=300 \
# Logging
LOG_LEVEL=INFO \
MCP_LOG_FILE=/app/logs/mcp.log
# Create API key configuration script
RUN echo '#!/bin/bash\n\
if [ -z "$VOYAGE_AI_API_KEY" ]; then\n\
echo "⚠️ WARNING: Semantic search requires VOYAGE_AI_API_KEY" >&2\n\
echo " Get your API key at: https://www.voyageai.com/" >&2\n\
echo " Set it using: -e VOYAGE_AI_API_KEY=your-key" >&2\n\
echo "" >&2\n\
echo " Costs: ~$0.05 per 1M tokens (1 token ≈ 1 character)" >&2\n\
echo " Free tier: 50M tokens/month (~50K embeddings)" >&2\n\
echo "" >&2\n\
echo " Falling back to BM25 search only..." >&2\n\
export SEMANTIC_SEARCH_ENABLED=false\n\
fi\n\
\n\
# Check for .env file\n\
if [ -f "/workspace/.env" ]; then\n\
echo "Loading configuration from .env file..." >&2\n\
export $(grep -v "^#" /workspace/.env | xargs)\n\
fi\n\
\n\
# Check GitHub artifact sync preferences\n\
if [ -f "/workspace/.mcp-index.json" ]; then\n\
echo "Loading artifact sync preferences..." >&2\n\
SYNC_ENABLED=$(python -c "import json; data=json.load(open(\"/workspace/.mcp-index.json\")); print(data.get(\"github_artifacts\", {}).get(\"enabled\", True))")\n\
export MCP_ARTIFACT_SYNC=$SYNC_ENABLED\n\
fi\n\
\n\
exec "$@"' > /app/docker-entrypoint.sh && \
chmod +x /app/docker-entrypoint.sh && \
chown mcp:mcp /app/docker-entrypoint.sh
# Switch back to non-root user
USER mcp
# Set the entrypoint to handle configuration
ENTRYPOINT ["/app/docker-entrypoint.sh", "python", "/app/mcp_server_cli.py"]
# Label for version tracking
LABEL org.opencontainers.image.title="MCP Index Server - Standard" \
org.opencontainers.image.description="AI-powered code search with semantic understanding" \
org.opencontainers.image.version="standard" \
org.opencontainers.image.source="https://github.com/Code-Index-MCP" \
org.opencontainers.image.documentation="https://github.com/Code-Index-MCP/docs/DOCKER_GUIDE.md" \
org.opencontainers.image.licenses="MIT" \
com.mcp.features="semantic-search,github-sync,reranking" \
com.mcp.api-keys="VOYAGE_AI_API_KEY"
