About this project

LLM Proxy is a local debugging tool that sits between applications and the OpenAI API to log and inspect traffic. It allows developers to monitor exact prompts, token usage, timing, and errors in real-time. Key capabilities include: - Request Logging: Captures all /v1/* endpoints, including streaming responses, and saves them to disk in JSONL format. - Web UI: Provides a browser-based interface to browse request history, filter by endpoint or model, and inspect full request/response bodies. - Configuration: Supports custom ports, API key overrides, and configurable logging levels via a config file or environment variables. - Security: Sanitizes API keys in logs (showing only the last 4 characters) and binds to localhost by default. - Integration: Compatible with any OpenAI-compatible client or SDK by changing the base URL to the local proxy address.