Configuration

Configure Lumen with lumen.toml.

Configuration File

Lumen looks for lumen.toml in:

1. Current directory
2. Parent directories (walking up)
3. ~/.config/lumen/lumen.toml

Package Configuration

[package]
name = "my-project"
version = "0.1.0"
description = "A Lumen project"
authors = ["Your Name <[email protected]>"]
license = "MIT"

Tool Providers

Map tools to implementations:

[providers]
llm.chat = "openai-compatible"
llm.embed = "openai-compatible"
http.get = "builtin-http"
http.post = "builtin-http"
postgres.query = "postgres"

OpenAI-Compatible

[providers.config.openai-compatible]
base_url = "https://api.openai.com/v1"
api_key_env = "OPENAI_API_KEY"
default_model = "gpt-4o"

# Optional
organization = "org-xxx"
timeout_ms = 30000
max_retries = 3

Ollama (Local)

[providers]
llm.chat = "openai-compatible"

[providers.config.openai-compatible]
base_url = "http://localhost:11434/v1"
default_model = "llama3"
# No API key needed for local

Anthropic

[providers]
llm.chat = "anthropic"

[providers.config.anthropic]
api_key_env = "ANTHROPIC_API_KEY"
default_model = "claude-3-opus-20240229"

HTTP Provider

[providers.config.builtin-http]
max_redirects = 5
timeout_ms = 30000
user_agent = "Lumen/1.0"

PostgreSQL

[providers.config.postgres]
connection_string_env = "DATABASE_URL"
pool_size = 10

MCP Servers

Configure Model Context Protocol servers:

[providers.mcp.github]
uri = "npx -y @modelcontextprotocol/server-github"
tools = ["github.create_issue", "github.search_repos"]
env = { GITHUB_TOKEN = "${GITHUB_TOKEN}" }

[providers.mcp.filesystem]
uri = "npx -y @modelcontextprotocol/server-filesystem /allowed/path"
tools = ["filesystem.read_file", "filesystem.write_file"]

MCP Configuration Options

Field	Description
uri	Command or URL to launch server
tools	List of tool names exposed
env	Environment variables
timeout_ms	Request timeout

Provider Types

Type	Description
openai-compatible	OpenAI, Ollama, vLLM, etc.
anthropic	Anthropic Claude API
builtin-http	Built-in HTTP client
postgres	PostgreSQL client
mcp	MCP server bridge

Secrets

Never put secrets directly in lumen.toml:

# BAD
api_key = "sk-xxx"  # Don't do this!

# GOOD
api_key_env = "OPENAI_API_KEY"  # Reference environment variable

Multiple Environments

Use environment-specific config:

# Development
LUMEN_ENV=dev lumen run app.lm.md

# Production
LUMEN_ENV=prod lumen run app.lm.md

With lumen.dev.toml and lumen.prod.toml.

Example: Full Configuration

[package]
name = "ai-assistant"
version = "1.0.0"

[providers]
llm.chat = "openai-compatible"
llm.embed = "openai-compatible"
http.get = "builtin-http"

[providers.config.openai-compatible]
base_url = "${OPENAI_BASE_URL:-https://api.openai.com/v1}"
api_key_env = "OPENAI_API_KEY"
default_model = "gpt-4o"
timeout_ms = 60000

[providers.config.builtin-http]
max_redirects = 3
timeout_ms = 10000

[providers.mcp.github]
uri = "npx -y @modelcontextprotocol/server-github"
tools = ["github.search_repos", "github.create_issue"]
env = { GITHUB_TOKEN = "${GITHUB_TOKEN}" }

[cache]
enabled = true
ttl_ms = 3600000  # 1 hour

[logging]
level = "info"
format = "json"

Next Steps

• CLI Reference — Command-line tools
• Providers — Provider details