Monitor Your Claude AI Usage Right from Your Mac Menu Bar
LunaRoute is a high-performance secure local proxy for AI coding assistants like Claude Code, Codex, and OpenCode. Get complete visibility into every LLM interaction with zero-overhead passthrough, session recording, and powerful debugging capabilities
LunaRoute detects and redacts PII (emails, SSNs, credit cards, phone numbers) before writing to disk using configurable redaction modes (mask, remove, tokenize, partial). Data stays local and never leaves the user's machine.
LunaRoute measures proxy overhead (typically 0.1-0.2ms) separately from provider latency, showing users whether slowness comes from their code or the LLM. Session statistics display average response times and tool call performance metrics.
LunaRoute accepts both OpenAI and Anthropic API formats simultaneously in passthrough mode, routing gpt-* models to OpenAI and claude-* models to Anthropic with zero normalization overhead. Users configure it once and both APIs work automatically.
LunaRoute records every LLM interaction (requests, responses, tokens, tool calls) with zero configuration, providing complete visibility into what Claude Code and other AI assistants send and receive. Users can search past sessions, inspect raw JSON, and analyze conversation patterns.
LunaRoute automatically breaks down input, output, and thinking tokens by session and model, calculates estimated costs in USD, and stores this data in SQLite for analytics. Users can query token usage patterns and identify expensive interactions.