Reference
Configuration
Every setting Bristlenose reads, with its default.
This page lists the settings the command line reads. In the macOS app, the same things are set in the Settings window (⌘,) — see the setup guides under How-to guides.
Bristlenose reads settings from four places. Later sources override earlier ones:
- Built-in defaults
- A
.envfile (searched upward from the working directory) BRISTLENOSE_*environment variables- Command-line flags
API keys follow a separate order: the system credential store first, then the matching environment variable, then .env. Bristlenose reads .env but never writes to it.
Provider settings
| Variable | Default | Meaning |
|---|---|---|
BRISTLENOSE_LLM_PROVIDER |
anthropic |
anthropic, openai, azure, google, or local. |
BRISTLENOSE_LLM_MODEL |
claude-sonnet-4-6 |
Model name for the active provider. |
BRISTLENOSE_ANTHROPIC_API_KEY |
— | Claude API key. |
BRISTLENOSE_OPENAI_API_KEY |
— | ChatGPT API key. |
BRISTLENOSE_GOOGLE_API_KEY |
— | Gemini API key. |
BRISTLENOSE_AZURE_API_KEY |
— | Azure OpenAI key. |
BRISTLENOSE_AZURE_ENDPOINT |
— | Azure resource endpoint URL. |
BRISTLENOSE_AZURE_DEPLOYMENT |
— | Azure deployment name (used in place of a model name). |
BRISTLENOSE_AZURE_API_VERSION |
2024-10-21 |
Azure API version. |
BRISTLENOSE_LOCAL_URL |
http://localhost:11434/v1 |
Ollama server address. |
BRISTLENOSE_LOCAL_MODEL |
llama3.2:3b |
Local model name. |
BRISTLENOSE_LLM_MAX_TOKENS |
64000 |
Maximum output tokens per call. |
BRISTLENOSE_LLM_TEMPERATURE |
0.1 |
Sampling temperature. |
BRISTLENOSE_LLM_CONCURRENCY |
3 |
Number of analysis calls run in parallel. |
Transcription settings
| Variable | Default | Meaning |
|---|---|---|
BRISTLENOSE_WHISPER_MODEL |
large-v3-turbo |
Transcription model size. Smaller is faster and less accurate. |
BRISTLENOSE_WHISPER_BACKEND |
auto |
auto, mlx, or faster-whisper. |
BRISTLENOSE_WHISPER_LANGUAGE |
en |
Spoken language of the recordings. |
Analysis and privacy settings
| Variable | Default | Meaning |
|---|---|---|
BRISTLENOSE_PII_ENABLED |
false |
Enable PII redaction (also set by --redact-pii). |
BRISTLENOSE_PII_SCORE_THRESHOLD |
0.7 |
Detection confidence for redaction, 0.0–1.0. |
BRISTLENOSE_MIN_QUOTE_WORDS |
5 |
Minimum words for a quote to be extracted. |
BRISTLENOSE_CODEBOOK |
— | Codebook framework to apply during analysis. |
BRISTLENOSE_COLOR_SCHEME |
auto |
Report theme: auto, light, or dark. |
BRISTLENOSE_PALETTE |
default |
Colour palette: default or edo. Also set by serve --palette. See Appearance. |
BRISTLENOSE_LOG_LEVEL |
INFO |
Log-file verbosity. Independent of -v. |
To store an API key rather than set it as an environment variable, use bristlenose configure <provider> — see the per-provider guides under How-to guides.