How do I install the trace-claude-code skill?

You can use the provided setup script in the repository or manually configure your .claude/settings.local.json file with your Braintrust API key and the hook script paths.

Does this work with any Braintrust project?

Yes, you can define your preferred project name using the BRAINTRUST_CC_PROJECT environment variable in your Claude configuration.

What dependencies are required for this skill?

You will need the Claude Code CLI, a Braintrust account and API key, and the 'jq' command-line tool installed on your machine.

Can I see individual terminal commands in the traces?

Yes, the PostToolUse hook captures every individual operation, including file reads, edits, and terminal commands like npm tests or git commits, as child spans.

What is Braintrust Claude Code tracing?

It is a specialized skill that captures your entire Claude Code session, including every terminal command and file edit, and sends it to Braintrust for visualization and analysis.

Trace Claude Code to Braintrust

Name: Trace Claude Code to Braintrust
Author: braintrustdata

bybraintrustdata

•

分析と監視

Automatically captures and logs hierarchical traces of Claude Code sessions, tool calls, and conversation turns to Braintrust for complete observability.

概要

This skill provides deep visibility into your AI coding workflows by automatically transmitting Claude Code interactions to the Braintrust platform. It structures your sessions into hierarchical traces, mapping out every user prompt, assistant response, and intermediate tool execution—such as file system operations and terminal commands. It is an essential tool for developers who need to debug complex AI interactions, audit automated code changes, or optimize their development process with data-driven insights.

主な機能

Quick-start setup script for rapid project-level configuration
Hierarchical trace capturing including sessions, turns, and tool calls
Seamless integration via Claude Code lifecycle hooks
Detailed metadata support for session IDs and workspace context
Automatic logging of terminal commands and file system operations
3 GitHub stars

ユースケース

Auditing AI-generated code changes and command executions for security and compliance
Analyzing LLM performance and response accuracy during interactive development
Debugging complex AI coding sessions where tool execution order and output are critical

概要

主な機能

Quick-start setup script for rapid project-level configuration
Hierarchical trace capturing including sessions, turns, and tool calls
Seamless integration via Claude Code lifecycle hooks
Detailed metadata support for session IDs and workspace context
Automatic logging of terminal commands and file system operations
3 GitHub stars

ユースケース

Auditing AI-generated code changes and command executions for security and compliance
Analyzing LLM performance and response accuracy during interactive development
Debugging complex AI coding sessions where tool execution order and output are critical