Configuration

Brief uses OpenAI-compatible APIs. Configuration can come from environment variables or ~/.config/brief/config.

Quick Start

Run brief — the app prompts for your API key and saves it. Done.

Or set it in your shell:

export OPENAI_API_KEY="sk-..."

Config File

Settings are stored in ~/.config/brief/config:

openai.api_key=sk-...
openai.base_url=https://api.openai.com/v1
model=gpt-4o
reasoning.effort=high
user.name=Your Name
config.priority=env

Environment Variables

Variable	Description
OPENAI_API_KEY	API key (required)
OPENAI_BASE_URL	Custom endpoint for alternative providers
LLM_MODEL	Default model ID
BRIEF_REASONING_EFFORT	Optional canonical Chat Completions reasoning_effort; accepted values are defined by ReasoningEffort
BRIEF_CONFIG_PRIORITY	env (default) or config — which source wins when both set

Reasoning Effort

Leave reasoning effort unset to preserve the selected model or provider default. Set it explicitly with BRIEF_REASONING_EFFORT, reasoning.effort in the config file, or brief --reasoning-effort <value>; the command-line choice takes precedence over the configured source.

Brief forwards every explicit value as the standard Chat Completions reasoning_effort field. It does not infer capabilities from the model name or base URL, and it does not silently change an unsupported value. An OpenAI-compatible provider can therefore accept or reject the requested field according to its own API support.

ReasoningEffort is the sole accepted-value owner.

Display Flags

Variable	Values	Description
BRIEF_ALT_SCREEN	1	Alternate screen buffer (clears on exit)
BRIEF_MOUSE	0/off/native/false, 1/all/true, wheel/btn/buttons, select	Mouse tracking mode (default: select)
BRIEF_SHOW_TOOLS	1	Show tool call messages

Default select captures the mouse wheel so chat history scrolls inside Brief and drag selection copies chat text directly. Set BRIEF_MOUSE=0 to keep all mouse behavior native to the terminal.

Ghostty on macOS and Linux supports using Shift to bypass application mouse reporting for terminal selection by default. If you prefer that workflow full-time, keep BRIEF_MOUSE=0.

Alternative Providers

OpenRouter

export OPENAI_API_KEY="sk-or-..."
export OPENAI_BASE_URL="https://openrouter.ai/api/v1"
export LLM_MODEL="anthropic/claude-sonnet-4-20250514"

Ollama (Local)

export OPENAI_API_KEY="ollama"
export OPENAI_BASE_URL="http://localhost:11434/v1"
export LLM_MODEL="llama3.2"

Requires: ollama serve and ollama pull llama3.2

LMStudio (Local)

export OPENAI_API_KEY="lm-studio"
export OPENAI_BASE_URL="http://localhost:1234/v1"

Development

For local development, copy .env-example to .env:

cp .env-example .env
make run

The .env file is gitignored and only used with make run.