Configuration

Settings live in config.toml (find it with kunji config path). Change them in three ways:

the menu: Settings
a command: kunji config set <key> <value>
editing config.toml in any text editor

API keys are never stored in this file.

kunji config show                          # everything
kunji config get display.currency          # one value
kunji config set display.currency INR      # change one
kunji config unset chat.system_prompt      # back to the default

All settings

Key	Default	What it does
`default_model`	—	Model used when you don’t pass `-m`, as `provider:model`
`fallback`	`[]`	Models to try when yours fails (guide)
`chat.system_prompt`	`""`	System prompt for new chats
`chat.max_tokens`	`4096`	Longest answer, in tokens
`chat.history`	`true`	Save chats on this computer (guide)
`display.currency`	`USD`	Currency for costs: `USD`, `INR`, `EUR`, `GBP`, `JPY` or any code
`display.usd_rate`	`1.0`	How many units of your currency per 1 US dollar
`display.max_width`	`100`	Widest that kunji draws content, in columns
`budget.daily.<provider>`	—	Daily limit for one provider, in your currency
`budget.daily.all`	—	Daily limit for all providers together
`updates.check`	`true`	Daily check for a new version
`network.catalog`	`true`	Daily download of public price lists
`cloud.url`	`""`	Optional kunji cloud server (off when empty)

Example config.toml

default_model = "anthropic:claude-sonnet-4-5"
fallback = ["groq:llama-3.3-70b-versatile", "ollama:llama3.2"]

[chat]
system_prompt = ""
max_tokens = 4096
history = true

[display]
currency = "INR"
usd_rate = 84
max_width = 100

[budget.daily]
all = 400
openai = 150

[updates]
check = true

[network]
catalog = true

[providers.openai]                  # written by `kunji keys add`
default_model = "gpt-4o-mini"
# base_url = "https://my-proxy.example.com/v1"

[custom_providers.lmstudio]         # written by `kunji providers add`
name = "LM Studio"
base_url = "http://localhost:1234/v1"
needs_key = false

[pricing."groq:llama-3.3-70b-versatile"]   # your own price, USD per 1M tokens
input = 0.59
output = 0.79

Values from the command line

kunji config set understands true/false, numbers and JSON lists:

kunji config set chat.history false
kunji config set chat.max_tokens 8192
kunji config set fallback '["groq:llama-3.3-70b-versatile"]'