Configuration

Environment variables and configuration options for the NAICS MCP Server.

Environment Variables

Server Settings

Variable	Default	Description
`NAICS_HOST`	`127.0.0.1`	Bind address
`NAICS_PORT`	`9090`	HTTP port
`NAICS_LOG_LEVEL`	`INFO`	Log level (`DEBUG`, `INFO`, `WARNING`, `ERROR`)
`NAICS_LOG_FORMAT`	`text`	Log format (`text`, `json`)

Data Settings

Variable	Default	Description
`NAICS_DATA_DIR`	`~/.naics-mcp`	Data directory
`NAICS_DB_PATH`	`{data_dir}/naics.db`	SQLite database path
`NAICS_EMBEDDINGS_PATH`	`{data_dir}/embeddings.npy`	Embeddings file

Search Settings

Variable	Default	Description
`NAICS_EMBEDDING_MODEL`	`all-MiniLM-L6-v2`	Sentence transformer model
`NAICS_DEFAULT_SEARCH_LIMIT`	`10`	Default search results limit
`NAICS_MIN_CONFIDENCE`	`0.3`	Minimum confidence threshold

Performance Settings

Variable	Default	Description
`NAICS_WORKERS`	`4`	Number of worker threads
`NAICS_MAX_BATCH_SIZE`	`100`	Maximum batch classification size
`NAICS_CACHE_SIZE`	`1000`	Search result cache size

Configuration File

Create a configuration file at ~/.naics-mcp/config.yaml:


# Server settings
server:
  host: "0.0.0.0"
  port: 9090
 
# Logging
logging:
  level: INFO
  format: json
 
# Search
search:
  default_limit: 10
  min_confidence: 0.3
  embedding_model: all-MiniLM-L6-v2
 
# Performance
performance:
  workers: 4
  max_batch_size: 100
  cache_size: 1000

CLI Options

The naics-mcp serve command accepts options that override configuration:


naics-mcp serve \
  --host 0.0.0.0 \
  --port 8080 \
  --log-level DEBUG

Available Options

Option	Description
`--host`	Bind address
`--port`	HTTP port
`--log-level`	Log level
`--data-dir`	Data directory
`--config`	Path to config file

Priority Order

Configuration is resolved in this order (highest to lowest priority):

Command-line options
Environment variables
Configuration file
Default values

Production Configuration

Docker


docker run -p 9090:9090 \
  -e NAICS_LOG_LEVEL=INFO \
  -e NAICS_LOG_FORMAT=json \
  -e NAICS_HOST=0.0.0.0 \
  ghcr.io/mfbaig35r/naics-mcp-server:latest

Kubernetes ConfigMap


apiVersion: v1
kind: ConfigMap
metadata:
  name: naics-mcp-config
data:
  NAICS_LOG_LEVEL: "INFO"
  NAICS_LOG_FORMAT: "json"
  NAICS_HOST: "0.0.0.0"
  NAICS_PORT: "9090"

Logging Configuration

Text Format (Development)


export NAICS_LOG_FORMAT=text
export NAICS_LOG_LEVEL=DEBUG

Output:


2024-01-15 10:30:00 INFO  [naics_mcp] Server starting on 0.0.0.0:9090
2024-01-15 10:30:01 DEBUG [naics_mcp.search] Loading embeddings...

JSON Format (Production)


export NAICS_LOG_FORMAT=json
export NAICS_LOG_LEVEL=INFO

Output:


{"timestamp":"2024-01-15T10:30:00Z","level":"INFO","logger":"naics_mcp","message":"Server starting on 0.0.0.0:9090"}

Model Configuration

Embedding Model

The default model is all-MiniLM-L6-v2. To use a different model:


export NAICS_EMBEDDING_MODEL=all-mpnet-base-v2
naics-mcp embeddings  # Regenerate embeddings
naics-mcp serve

Supported models:

all-MiniLM-L6-v2 (default, fast, 384 dimensions)
all-mpnet-base-v2 (slower, better quality, 768 dimensions)
Any sentence-transformers compatible model

Regenerating Embeddings

After changing the model:


# Remove existing embeddings
rm ~/.naics-mcp/embeddings.npy
 
# Generate new embeddings
naics-mcp embeddings
 
# Restart server
naics-mcp serve

Security Configuration

Bind to Localhost Only

For development or when behind a reverse proxy:


export NAICS_HOST=127.0.0.1

Bind to All Interfaces

For container deployments:


export NAICS_HOST=0.0.0.0

Rate Limiting

Configure rate limits (future feature):


rate_limiting:
  enabled: true
  requests_per_minute: 60
  burst: 10

Troubleshooting

Check Current Configuration


naics-mcp config show

Validate Configuration


naics-mcp config validate

Reset to Defaults


rm ~/.naics-mcp/config.yaml
naics-mcp init