Skip to content

configuration.md

Latest commit

 

History

History
136 lines (101 loc) · 4.43 KB

configuration.md

File metadata and controls

136 lines (101 loc) · 4.43 KB

Configuration

duh uses TOML configuration files with a layered merge system. You can configure providers, cost limits, consensus behavior, and more.

Config file locations

Configuration is loaded in this order (later overrides earlier):

  1. Built-in defaults -- Sensible defaults from the Pydantic schema
  2. User config -- ~/.config/duh/config.toml (or $XDG_CONFIG_HOME/duh/config.toml)
  3. Project config -- ./duh.toml in the current directory
  4. $DUH_CONFIG -- Explicit path via environment variable
  5. CLI --config -- Path passed to any command
  6. Programmatic overrides -- When using duh as a library

Each layer deep-merges into the previous. You only need to specify the values you want to change.

Minimal config

Create ~/.config/duh/config.toml:

[providers.anthropic]
api_key_env = "ANTHROPIC_API_KEY"

[providers.openai]
api_key_env = "OPENAI_API_KEY"

!!! tip "This is the default" duh ships with this exact config as defaults. If your API keys are set as environment variables, you don't need a config file at all.

Full config example

[general]
max_rounds = 3           # Maximum consensus rounds (1-10)
stream_output = true     # Stream model responses
protocol = "consensus"   # Default protocol: consensus, voting, or auto
decompose = false        # Decompose questions into subtasks by default

[database]
url = "sqlite+aiosqlite:///~/.local/share/duh/duh.db"

[cost]
warn_threshold = 1.00    # Warn when cumulative cost exceeds this (USD)
hard_limit = 10.00       # Stop when cumulative cost exceeds this (USD)
show_running_cost = true # Display cost after each round

[providers.anthropic]
enabled = true
api_key_env = "ANTHROPIC_API_KEY"
# api_key = "sk-ant-..."   # Or set directly (not recommended)

[providers.openai]
enabled = true
api_key_env = "OPENAI_API_KEY"
# base_url = "http://localhost:11434/v1"  # For Ollama / local models

[providers.google]
enabled = true
api_key_env = "GOOGLE_API_KEY"

[consensus]
min_challengers = 2
proposer_strategy = "round_robin"
challenge_types = ["flaw", "alternative", "risk", "devils_advocate"]
# panel = ["anthropic:claude-opus-4-6", "openai:gpt-5.2"]  # Restrict to specific models

[tools]
enabled = false          # Enable tool-augmented reasoning
max_rounds = 5           # Max tool call iterations per phase

[tools.web_search]
backend = "duckduckgo"   # Search backend
max_results = 5          # Max results per search

[tools.code_execution]
enabled = false          # Code execution requires explicit opt-in
timeout = 30             # Execution timeout in seconds
max_output = 10000       # Max output characters

[voting]
enabled = false          # Enable voting protocol availability
aggregation = "majority" # Aggregation strategy: majority or weighted

[decompose]
max_subtasks = 7         # Maximum subtasks per decomposition
parallel = true          # Execute independent subtasks in parallel

[taxonomy]
enabled = false          # Auto-classify decisions by intent/category/genus
model_ref = ""           # Model for classification (empty = cheapest)

[logging]
level = "INFO"
structured = false

Environment variables

Provider API keys are resolved from environment variables when api_key_env is set and api_key is not provided directly.

Variable Purpose
ANTHROPIC_API_KEY Anthropic (Claude) API key
OPENAI_API_KEY OpenAI API key
GOOGLE_API_KEY Google (Gemini) API key
DUH_CONFIG Path to config file (overrides discovery)
XDG_CONFIG_HOME Base directory for user config (default: ~/.config)

Override config per command

Every command accepts --config to use a specific config file:

duh --config ./my-config.toml ask "question"

The ask command accepts several overrides:

duh ask --rounds 5 "question requiring deeper debate"
duh ask --protocol voting "quick judgment call"
duh ask --decompose "complex multi-part question"
duh ask --tools "question needing web search"
duh ask --no-tools "question that should not use tools"
duh ask --panel anthropic:claude-opus-4-6,openai:gpt-5.2 "restrict to specific models"
duh ask --proposer openai:gpt-5.2 "use a specific proposer"
duh ask --challengers google:gemini-3-pro-preview,perplexity:sonar-pro "specific challengers"

Next steps