Implementation:Openai Openai node Logger

Knowledge Sources	Openai_Openai_node
Domains	SDK, Logging
Last Updated	2026-02-15 12:00 GMT

Overview

The Logger module provides the SDK's internal logging system with configurable log levels and request detail formatting.

Description

The logging system defines a Logger interface with four severity methods (error, warn, info, debug) and a LogLevel type that controls which messages are emitted. Log levels are ranked numerically: off (0), error (200), warn (300), info (400), and debug (500). When a log function's level exceeds the configured threshold, it is replaced with a no-op function, ensuring zero overhead for suppressed messages.

The loggerFor function retrieves or constructs a level-filtered logger for a given OpenAI client instance. It uses a WeakMap cache keyed on the underlying logger object to avoid repeatedly constructing filtered loggers. If no logger is configured, a no-op logger is returned.

The module also exports formatRequestDetails, which sanitizes request metadata for logging by redacting sensitive headers (Authorization, Cookie, Set-Cookie) and cleaning up redundant fields. The parseLogLevel function validates string-based log level inputs, warning if an unrecognized value is provided.

Usage

Use the logging utilities when you need to add debug or diagnostic output to SDK operations. Configure a logger on the OpenAI client instance and set the desired logLevel. The formatRequestDetails function should be used when logging HTTP request and response details to ensure sensitive information is not leaked.

Code Reference

Source Location

• Repository: openai-node
• File: src/internal/utils/log.ts

Signature

type LogFn = (message: string, ...rest: unknown[]) => void;

export type Logger = {
  error: LogFn;
  warn: LogFn;
  info: LogFn;
  debug: LogFn;
};

export type LogLevel = 'off' | 'error' | 'warn' | 'info' | 'debug';

export const parseLogLevel: (
  maybeLevel: string | undefined,
  sourceName: string,
  client: OpenAI,
) => LogLevel | undefined;

export function loggerFor(client: OpenAI): Logger;

export const formatRequestDetails: (details: {
  options?: RequestOptions | undefined;
  headers?: Headers | Record<string, string> | undefined;
  retryOfRequestLogID?: string | undefined;
  retryOf?: string | undefined;
  url?: string | undefined;
  status?: number | undefined;
  method?: string | undefined;
  durationMs?: number | undefined;
  message?: unknown;
  body?: unknown;
}) => object;

Import

import { loggerFor, formatRequestDetails, parseLogLevel } from 'openai/internal/utils/log';
import type { Logger, LogLevel } from 'openai/internal/utils/log';

I/O Contract

Inputs

Name	Type	Required	Description
client	OpenAI	Yes	The OpenAI client instance whose logger and logLevel properties are read to construct a filtered logger.
details	object	Yes (for formatRequestDetails)	Request metadata object containing headers, URL, status, method, duration, and other fields to be sanitized for logging.
maybeLevel	undefined	Yes (for parseLogLevel)	A candidate log level string to validate against known levels.

Outputs

Name	Type	Description
Logger	Logger	A logger object with methods filtered by the configured log level. Methods below the threshold are no-ops.
sanitized details	object	Request details with sensitive headers redacted and redundant fields cleaned up.

Usage Examples

import OpenAI from 'openai';

const client = new OpenAI({
  logger: console,
  logLevel: 'debug',
});

// The SDK internally uses loggerFor to get a filtered logger:
// import { loggerFor } from 'openai/internal/utils/log';
// const log = loggerFor(client);
// log.debug('Request started', formatRequestDetails({ url, method, headers }));

import { formatRequestDetails } from 'openai/internal/utils/log';

const sanitized = formatRequestDetails({
  url: 'https://api.openai.com/v1/chat/completions',
  method: 'POST',
  status: 200,
  durationMs: 342,
  headers: {
    'Authorization': 'Bearer sk-...',
    'Content-Type': 'application/json',
  },
});
// sanitized.headers['Authorization'] === '***'
// sanitized.headers['Content-Type'] === 'application/json'

Related Pages

• Environment:Openai_Openai_node_Node_20_Runtime