CodingAgentExplorer is an open-source tool to explore and visualize how Claude Code works under the hood. Run it as a reverse proxy and live dashboard to inspect prompts, tool calls, hooks, MCP traffic, token usage, and the full agent conversation loop in real time. Built for developers who want to understand, debug, and teach agentic development.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

Ever wondered what actually happens when you ask Claude Code to fix a bug? 🤔 * What files does it read? * Which tools does it call? * What prompts does it send to the model? * How do MCP tools and hooks fit into the process? I built CodingAgentExplorer to answer those questions. It’s an open-source tool that sits between Claude Code and the API and lets you watch everything in real time: * prompts and responses * tool calls (Read, Write, Bash, etc.) * hook events * MCP traffic * token usage and timing * the full agent conversation loop I originally created it as a teaching tool for workshops on agentic development, because many developers feel coding agents are still a black box. The goal is simple: Make AI coding agents observable and understandable. I’d love feedback from the community: What insights would you want to see when an agent runs? Are there other agents or MCP tools you'd like supported? GitHub contributions and ideas are very welcome!