Configuration

You can use the following ways to configure Logfire:

1. Programmatically via logfire.configure()
2. Using environment variables
3. Using a configuration file (pyproject.toml)

The order of precedence is as above.

Programmatically via configure

For more details, please refer to our API documentation.

Using environment variables

You can use the following environment variables to configure Logfire:

Name	Description
LOGFIRE_BASE_URL	The base URL of the Logfire backend. Primarily for testing purposes.
LOGFIRE_SEND_TO_LOGFIRE	Whether to send spans to Logfire.
LOGFIRE_MIN_LEVEL	Minimum log level for logs and spans to be created. By default, all logs and spans are created.
LOGFIRE_TOKEN	Token for the Logfire API.
LOGFIRE_SERVICE_NAME	Name of the service emitting spans. For further details, please refer to the Service section.
LOGFIRE_SERVICE_VERSION	Version number of the service emitting spans. For further details, please refer to the Service section.
LOGFIRE_ENVIRONMENT	Environment in which the service is running. For further details, please refer to the Deployment section.
LOGFIRE_TRACE_SAMPLE_RATE	Head sampling rate for traces.
LOGFIRE_CREDENTIALS_DIR	The directory where to store the configuration file.
LOGFIRE_CONSOLE	Whether to enable/disable the console exporter.
LOGFIRE_CONSOLE_COLORS	Whether to use colors in the console.
LOGFIRE_CONSOLE_SPAN_STYLE	How spans are shown in the console.
LOGFIRE_CONSOLE_INCLUDE_TIMESTAMP	Whether to include the timestamp in the console.
LOGFIRE_CONSOLE_INCLUDE_TAGS	Whether to include tags in the console.
LOGFIRE_CONSOLE_VERBOSE	Whether to log in verbose mode in the console.
LOGFIRE_CONSOLE_MIN_LOG_LEVEL	Minimum log level to show in the console.
LOGFIRE_CONSOLE_SHOW_PROJECT_LINK	Whether to enable/disable the console exporter.
LOGFIRE_PYDANTIC_PLUGIN_RECORD	Whether instrument Pydantic validation..
LOGFIRE_PYDANTIC_PLUGIN_INCLUDE	Set of items that should be included in Logfire Pydantic plugin instrumentation.
LOGFIRE_PYDANTIC_PLUGIN_EXCLUDE	Set of items that should be excluded from Logfire Pydantic plugin instrumentation.
LOGFIRE_INSPECT_ARGUMENTS	Whether to enable the f-string magic feature. On by default for Python 3.11 and above.
LOGFIRE_IGNORE_NO_CONFIG	Whether to show a warning message if logfire if used without calling logfire.configure()
LOGFIRE_DISTRIBUTED_TRACING	Whether to extract incoming trace context. By default, will extract but warn about it.
LOGFIRE_HTTPX_CAPTURE_ALL	Whether to capture all HTTP headers, request and response bodies when using logfire.instrument_httpx()

When using environment variables, you still need to call logfire.configure(), but you can leave out the arguments.

Using a configuration file (pyproject.toml)

You can use the pyproject.toml to configure Logfire.

Here's an example:

[tool.logfire]
project_name = "My Project"
console_colors = "never"

The keys are the same as the parameters of logfire.configure().