promptcmd

Appearance

Configuration File ​

promptcmd uses a single TOML configuration file to manage all settings, including provider credentials, global parameters, variants, and load balancing groups.

File Location ​

You can create/edit your configuration simply via:

bash
promptctl config edit

Or by manually creating/editing the file.

Linux ​

~/.config/promptcmd/config.toml

macOS ​

~/Library/Application Support/promptcmd/config.toml

Windows ​

%APPDATA%\promptcmd\config.toml

Basic Structure ​

A minimal configuration file looks like:

toml

[providers.anthropic]
api_key = "sk-ant-api03-..."

Global Provider Settings ​

Settings that apply to all providers unless overridden:

toml
[providers]
temperature = 0.7       # Sampling temperature (0.0 - 1.0)
max_tokens = 1000       # Maximum tokens in response
cache_ttl = 20          # Cache time-to-live in seconds

Temperature ​

Controls randomness in responses:

  • 0.0: Deterministic, focused responses
  • 0.5: Balanced creativity and consistency
  • 1.0: Maximum creativity and randomness
toml
[providers]
temperature = 0.3  # Lower for code, higher for creative writing

Max Tokens ​

Maximum length of model responses:

toml
[providers]
max_tokens = 2000  # Adjust based on your needs

Cache TTL ​

How long to cache identical requests (in seconds):

toml
[providers]
cache_ttl = 60  # Cache for 60 seconds

Set to 0 to disable caching.

Provider-specific Configuration ​

Configure individual AI providers with their specific settings.

Anthropic (Claude) ​

toml
[providers.anthropic]
api_key = "sk-ant-api03-..."

Optional settings:

toml
[providers.anthropic]
api_key = "sk-ant-api03-..."
endpoint = "https://api.anthropic.com"  # Custom endpoint
temperature = 0.5                        # Override global
max_tokens = 4000                       # Override global
cache_ttl = 30                          # Override global

OpenAI (GPT Models) ​

toml
[providers.openai]
api_key = "sk-..."
endpoint = "https://api.openai.com/v1"

Google (Gemini) ​

toml
[providers.google]
api_key = "..."

OpenRouter ​

toml
[providers.openrouter]
api_key = "sk-or-..."
endpoint = "https://openrouter.ai/api/v1"

Ollama (Local Models) ​

toml
[providers.ollama]
endpoint = "http://localhost:11434"

For Ollama, no API key is needed since it runs locally.

Configuration Hierarchy ​

Settings cascade from general to specific:

toml
# Global defaults
[providers]
temperature = 0.7
max_tokens = 1000
cache_ttl = 20

# Provider-specific overrides
[providers.anthropic]
api_key = "sk-ant-..."
temperature = 0.5      # Overrides global for Anthropic

# Variant-specific overrides
[providers.anthropic.fast-responder]
temperature = 0.3      # Overrides both global and provider
max_tokens = 500
system = "Be extremely brief."

Priority order (highest to lowest):

  1. Variant settings (most specific)
  2. Provider settings
  3. Global settings (least specific)