A CLI tool that turns AI coding agent terminal output into a navigable trace tree
AI coding agents like Claude Code and Cursor emit dense verbose JSON logs that are nearly impossible to read in real time, making it hard to understand what the agent actually did, which tool calls succeeded or failed, and where tokens were burned. Developers are left guessing why a run went wrong or what the agent changed. This tool parses the raw output stream in real time, reconstructs a structured trace tree of every tool call, file diff, subagent spawn, and token spend, and renders it as a navigable terminal or local-web UI that makes agent behavior immediately legible.
Demand Breakdown
Social Proof 2 sources
Gap Assessment
4 tools exist (LangSmith, Langfuse, claude-devtools, Braintrust) but gaps remain: Built for application developers instrumenting their own LLM pipelines, not for developers using Claude Code or Cursor as a black-box agent. No real-time terminal-side parsing of agent stdout; no tool-call tree reconstruction from raw CLI output; no subagent loop visualization.; Same gap as LangSmith: requires SDK instrumentation in your own code. Cannot parse Claude Code or Cursor raw terminal output. No CLI-native experience, no agent stdout interceptor..
Features8 agent-ready prompts
Competitive LandscapeFREE
| Product | Does | Missing |
|---|---|---|
| LangSmith | LLM trace observability and evaluation for production AI apps; logs prompts, completions, and chain steps; added a CLI fetch tool in late 2025 for pulling traces into coding agents. | Built for application developers instrumenting their own LLM pipelines, not for developers using Claude Code or Cursor as a black-box agent. No real-time terminal-side parsing of agent stdout; no tool-call tree reconstruction from raw CLI output; no subagent loop visualization. |
| Langfuse | Open-source LLM observability with prompt management, trace logging, and evaluation. Self-hostable. | Same gap as LangSmith: requires SDK instrumentation in your own code. Cannot parse Claude Code or Cursor raw terminal output. No CLI-native experience, no agent stdout interceptor. |
| claude-devtools | Early-stage visual debugger for Claude Code showing tool calls, token attribution, and subagent trees built by an indie developer. | Early-stage, no active distribution or monetization, no timeline replay, no diff viewer, no failure root-cause analysis, no cross-session comparison. Proof that the gap exists, not a solution that fills it. |
| Braintrust | AI evaluation and observability platform covering evals, logging, and prompt experiments for teams building AI products. | Enterprise/team focus on AI product developers, not individual developers debugging a coding agent CLI session. No stdout intercept, no coding-agent-specific trace reconstruction. |
Leads88BUILDER
Sign in to unlock full access.