kb-mcp-server

#!/bin/bash set -e # Default values PORT=${PORT:-8000} HOST=${HOST:-0.0.0.0} TRANSPORT=${TRANSPORT:-sse} EMBEDDINGS_PATH=${EMBEDDINGS_PATH:-/data/embeddings} CONFIG_FILE=${CONFIG_FILE:-config.yml} # Handle embeddings path if [ -n "$EMBEDDINGS_PATH" ]; then # Check if the embeddings path is a tar.gz file if [[ "$EMBEDDINGS_PATH" == *.tar.gz ]]; then echo "Detected tar.gz embeddings file: $EMBEDDINGS_PATH" # Pass the path directly to the embeddings parameter EMBEDDINGS_ARG="--embeddings $EMBEDDINGS_PATH" else echo "Using embeddings directory: $EMBEDDINGS_PATH" # Set environment variable for directory-based embeddings export TXTAI_EMBEDDINGS=$EMBEDDINGS_PATH EMBEDDINGS_ARG="--embeddings $EMBEDDINGS_PATH" fi else EMBEDDINGS_ARG="" fi # Handle config file if it exists and is not empty CONFIG_ARG="" if [ -n "$CONFIG_FILE" ] && [ -f "$CONFIG_FILE" ]; then echo "Using config file: $CONFIG_FILE" CONFIG_ARG="--config $CONFIG_FILE" fi # Print configuration echo "Starting TxtAI MCP Server with:" echo " - Transport: $TRANSPORT" echo " - Host: $HOST" echo " - Port: $PORT" echo " - Embeddings: $EMBEDDINGS_PATH" if [ -n "$CONFIG_ARG" ]; then echo " - Config: $CONFIG_FILE" fi # Print pre-cached models if any if [ -n "$HF_TRANSFORMERS_MODELS" ]; then echo " - Pre-cached Transformer models: $HF_TRANSFORMERS_MODELS" fi if [ -n "$HF_SENTENCE_TRANSFORMERS_MODELS" ]; then echo " - Pre-cached Sentence Transformer models: $HF_SENTENCE_TRANSFORMERS_MODELS" fi # Run the server with the specified parameters exec python -m txtai_mcp_server --transport "$TRANSPORT" --host "$HOST" --port "$PORT" $EMBEDDINGS_ARG $CONFIG_ARG