Do I need a Braintrust account to use this skill?

Yes, you need a Braintrust API key and a project set up on the Braintrust platform to view and store your traces.

Does this capture terminal output?

Yes, any tool call executed by Claude Code, including terminal commands and their results, is captured as a child span within the hierarchical trace.

Can I customize which projects are traced?

Yes, you can configure the Braintrust project name and enable/disable tracing via the .claude/settings.local.json file in your project directory.

How does the tracing affect performance?

The tracing is handled via lightweight shell hooks that execute alongside tool uses, resulting in minimal impact on the perceived speed of Claude Code.

What exactly does Braintrust Tracing capture?

It captures the entire Claude Code lifecycle, including the root session, individual conversation turns, and specific tool calls like file edits, reads, and terminal commands.

Braintrust Tracing for Claude Code

Name: Braintrust Tracing for Claude Code
Author: parcadei

byparcadei

•

1,498

分析と監視

Automatically captures and logs Claude Code session traces to Braintrust for full observability and debugging.

概要

This skill provides deep visibility into your Claude Code interactions by automatically exporting hierarchical traces to Braintrust. It captures every aspect of a session—including user prompts, assistant responses, and individual tool executions like file edits or terminal commands—organizing them into a structured tree. This is essential for developers who need to debug agentic workflows, audit AI-driven changes, or monitor the performance and reliability of Claude Code within their development environment using professional-grade observability tools.

主な機能

1,498 GitHub stars
Detailed metadata support including session IDs and workspace context
Automatic hierarchical tracing of sessions, turns, and tool calls
Real-time capture of file edits, reads, and terminal command outputs
Hook-based architecture for automated, non-intrusive data collection
Seamless integration with Braintrust observability dashboards

ユースケース

Monitoring AI coding performance and session reliability across a team
Auditing AI-suggested changes and terminal execution history for safety
Debugging complex agentic coding tasks involving multiple tool iterations

概要

主な機能

1,498 GitHub stars
Detailed metadata support including session IDs and workspace context
Automatic hierarchical tracing of sessions, turns, and tool calls
Real-time capture of file edits, reads, and terminal command outputs
Hook-based architecture for automated, non-intrusive data collection
Seamless integration with Braintrust observability dashboards

ユースケース

Monitoring AI coding performance and session reliability across a team
Auditing AI-suggested changes and terminal execution history for safety
Debugging complex agentic coding tasks involving multiple tool iterations