Intercept AI agent traffic for debugging. See link below 👇

Debugging coding agents is hard when the context is invisible. claude-tap is a local proxy and trace viewer for AI coding agents like Claude Code, Codex CLI, Gemini CLI, Cursor CLI, OpenCode, Kimi,…

🧵 1/2 Stop guessing what your coding agents are actually sending to the LLM. Most developers building agentic workflows hit a wall when an agent stalls or hallucinates, only to realize they have…

liaohch3/claude-tap: Intercept and inspect Coding Agent API traffic from Claude Code, Codex CLI... - Open-source GitHub repository - It has 1,618 stars and recent activity

平时用 Claude Code 或者 Cursor CLI 这些命令行工具。 有时候想知道它底层到底给模型发了什么提示词，或者具体消耗了多少 Token。 有个 Python 写的流量分析工具挺实用。 它可以追踪这些 AI 编程 CLI 的 API 请求。 每次调用的 Prompt、对话历史、工具调用结果都能直接看到。 比较方便的是，它能对比相邻的两次请求。 很容易定位到 Prompt…

Coding agents are a black box. Not anymore. Claude-tap is a local proxy for AI coding agents: - Inspect prompts, tool schemas, streaming responses - Works with Claude Code, Codex CLI, Gemini, Cursor…

Intercept and inspect Coding Agent API traffic from Claude Code, Codex CLI, Gemini CLI, Cursor CLI, OpenCode, Kimi, Pi, and Hermes in a local trace viewer. Where would this fit in your stack?