MCP Log Proxy

Overview

solves the common developer pain of inspecting the low‑level message traffic that flows between an MCP client and a server. In many AI assistant workflows, troubleshooting or debugging requires visibility into the JSON payloads exchanged over STDIO. Without a dedicated tool, developers must rely on verbose logging from the server itself or manual packet captures, both of which can be noisy and hard to correlate. bridges this gap by acting as a transparent middleman that captures, stores, and displays every request and response in real time through an intuitive web interface.

At its core, the proxy launches the target MCP server via a user‑supplied command line and then forwards all STDIO traffic through an HTTP endpoint. The web UI presents a chronological view of each message, highlighting request/response pairs and allowing developers to drill into individual fields. The interface also surfaces any errors that occur within the proxy itself, giving a single pane of glass for both application and transport diagnostics. Because it operates over STDIO only, the proxy is lightweight and compatible with any MCP implementation that follows the standard.

Key features include:

• Multi‑instance management – When several proxies run concurrently, each registers itself in a shared JSON file, and the UI offers a selector to switch between instances without restarting services.
• Customizable presentation – Flags for title, port, and log file paths let teams tailor the proxy to their deployment conventions.
• Error visibility – Separate logs capture proxy‑level failures, ensuring that issues in the intermediary are not hidden behind the server’s own logs.
• Zero‑configuration for many servers – Simply point the proxy at a known MCP binary; no additional instrumentation is required.

Typical use cases span from local development to CI pipelines. A data scientist can spin up alongside a prototype MCP server, then watch the conversation in their browser to validate that prompts and responses adhere to protocol expectations. In a production setting, operators can deploy the proxy as an audit layer, recording traffic for compliance or post‑mortem analysis without modifying the underlying AI assistant. Because the proxy communicates purely over HTTP, it integrates seamlessly into existing monitoring dashboards or alerting systems.

Overall, provides a focused, user‑friendly window into the heart of MCP interactions. By exposing every message in context and supporting multiple simultaneous instances, it empowers developers to debug, validate, and optimize AI assistant workflows with minimal friction.