config.yaml

# Xantus Configuration File
# Configure your LLM, embeddings, vector store, and RAG settings

# ===== LLM Configuration =====
llm:
  # Provider: ollama, openai, or anthropic
  provider: anthropic

  # Model name (depends on provider)
  # Ollama: llama3.2, mistral, etc.
  # OpenAI: gpt-4, gpt-3.5-turbo, etc.
  # Anthropic: claude-sonnet-4-20250514, claude-3-5-sonnet-20241022, etc.
  model: claude-sonnet-4-20250514

  # Temperature for sampling (0.0 to 2.0)
  temperature: 0.7

  # Maximum tokens to generate
  max_tokens: 2048

  # API key (required for OpenAI and Anthropic)
  # Can also be set via environment variable: XANTUS_LLM_API_KEY
  api_key: null

  # API base URL (optional, for custom endpoints)
  # For Ollama, defaults to http://localhost:11434
  api_base: null

# ===== Embedding Configuration =====
embedding:
  # Provider: huggingface, ollama, or openai
  provider: huggingface

  # Model name
  # HuggingFace: BAAI/bge-small-en-v1.5, sentence-transformers/all-MiniLM-L6-v2
  # Ollama: nomic-embed-text, mxbai-embed-large
  # OpenAI: text-embedding-3-small, text-embedding-ada-002
  model: BAAI/bge-small-en-v1.5

  # Batch size for embedding generation
  embed_batch_size: 10

  # API key (required for OpenAI)
  api_key: null

# ===== Vector Store Configuration =====
vector_store:
  # Provider: chroma or qdrant
  provider: chroma

  # Path to persist vector store data
  persist_path: ./data/vector_store

  # Collection name
  collection_name: xantus_documents

# ===== RAG Configuration =====
rag:
  # Number of similar chunks to retrieve
  similarity_top_k: 5

  # Size of text chunks for indexing (in characters)
  chunk_size: 1024

  # Overlap between chunks (in characters)
  chunk_overlap: 200

  # Enable reranking of retrieved chunks (requires additional setup)
  enable_reranking: false

# ===== Server Configuration =====
server:
  # Host to bind the server to
  host: 127.0.0.1

  # Port to bind the server to
  port: 8000

  # Enable CORS
  cors_enabled: true

  # Allowed CORS origins
  cors_origins:
    - "*"

# ===== Example Configurations =====

# Example 1: Local setup with Ollama
# llm:
#   provider: ollama
#   model: llama3.2
#   api_base: http://localhost:11434
# embedding:
#   provider: ollama
#   model: nomic-embed-text

# Example 2: Cloud setup with OpenAI
# llm:
#   provider: openai
#   model: gpt-4
#   api_key: sk-your-api-key-here
# embedding:
#   provider: openai
#   model: text-embedding-3-small
#   api_key: sk-your-api-key-here

# Example 3: Hybrid setup - Anthropic LLM with local embeddings
# llm:
#   provider: anthropic
#   model: claude-3-5-sonnet-20241022
#   api_key: sk-ant-your-api-key-here
# embedding:
#   provider: huggingface
#   model: BAAI/bge-small-en-v1.5

# Example 4: All local with Ollama
# llm:
#   provider: ollama
#   model: mistral
# embedding:
#   provider: ollama
#   model: nomic-embed-text
# vector_store:
#   provider: chroma
#   persist_path: ./data/vector_store

# ===== MCP Configuration =====
# Enable MCP to connect external tools (calculator, file system, etc.)
mcp:
  # Set to true to enable MCP integration
  enabled: true

  # List of MCP servers to connect to
  servers:
    - name: "mcp-starter-template"
      command: "node"
      args: ["mcp-servers/mcp-starter-template-ts/dist/start.js"]

  # Example: Using absolute path (for external MCP servers)
  # servers:
  #   - name: "external-mcp-server"
  #     command: "node"
  #     args: ["/absolute/path/to/server/index.js"]

  # Example: Multiple MCP servers
  # servers:
  #   - name: "my-tools"
  #     command: "node"
  #     args: ["/path/to/mcp-server/index.js"]
  #   - name: "database-tools"
  #     command: "npx"
  #     args: ["-y", "@modelcontextprotocol/server-postgres", "postgresql://localhost/mydb"]