Coding Agent Inspector Dashboard

Developers using AI coding agents (Claude, Codex, Kimi) can't see what data is being sent to the model. This tool provides a local proxy + web dashboard to inspect all requests/responses, helping developers debug, optimize prompts, and understand API usage patterns.

AI coding agent adoption is accelerating sharply in 2025, and as teams move from casual Copilot use to agentic workflows with multi-step tool calls, the opacity of what's actually hitting the API becomes a real debugging and cost-control problem. No clear incumbent owns this specific proxy-inspection layer for LLM traffic, though tools like Helicone and LangSmith address adjacent observability needs at the platform level rather than the local developer workflow level. The $500–2k/mo revenue band is realistic only if priced as a team or per-seat SaaS tool, since individual developers will expect a free tier and resist paying for something they could approximate with mitmproxy and a JSON viewer. The most likely failure mode is that the major AI coding agent providers — Anthropic, OpenAI — add native request inspection to their own developer consoles, making a third-party proxy redundant before any meaningful user base is established.