Braintrustdata Braintrust Claude Plugin
braintrustdata-braintrust-claude-plugin is a Claude Code plugin bundle for the Ship phase that connects Braintrust evaluation, conversation logging, and LLM observability to your local Claude Code sessions.
by braintrustdata · github.com/braintrustdata/braintrust-claude-plugin
Wire Claude Code agent runs into Braintrust for eval datasets, trace logging, and production-style LLM observability without leaving the editor.
Add it to Claude Code
Install the plugin in Claude Code. One command, paste-ready.
/plugin install braintrustdata-braintrust-claude-plugin@braintrustdata/braintrust-claude-pluginBuilt to be called by your agent
Skillselion is itself an MCP server. Your agent can pull this entry and a paste-ready install config straight from the API - no copy-paste.
Retrieve this entry with skillselion.get_details("plugin:braintrustdata/braintrust-claude-plugin") and the paste-ready config with skillselion.get_install_config("plugin:braintrustdata/braintrust-claude-plugin").
What it does
braintrustdata-braintrust-claude-plugin is a two-plugin Claude Code bundle from Braintrust that connects your local agent workflow to evaluation, structured logging, and observability. Solo builders shipping AI features—chat assistants, coding agents, or API-backed LLM products—install it when they need more than printf debugging: they want traces of what Claude said, which tools fired, and how new prompts compare to baselines. The marketplace listing emphasizes automatic capture of conversations and usage, plus evaluation-oriented helpers so you can judge whether changes actually improve outcomes. It is not a replacement for your application’s unit tests; it is the layer that makes LLM-specific quality measurable. Use it during Ship when you are hardening agent behavior, and keep it on during Operate when you need ongoing visibility into real coding sessions and regressions.
Highlights
- Bundle of 2 Claude Code plugins aimed at Braintrust eval, logging, and observability
- Automatically captures Claude Code conversations, messages, tool calls, and usage for traceability
- Helper scripts and examples for evaluation workflows against agent responses
- Enables observability across working sessions—not a one-off test script
- Fits agents and assistant-heavy builds that need logged, correctable LLM behavior
Why builders use it
You cannot tell whether prompt or tool changes actually improved your agent when Claude Code runs leave no structured traces or scored eval history.
After you register the bundle, agent conversations and calls can be logged and evaluated in Braintrust so you can compare versions and monitor behavior over time.
At a glance
- Type - Plugin in LLM Integration.
- Adoption - 0 installs, 16 stars, 0 votes.
FAQ
Who is braintrustdata-braintrust-claude-plugin for?
Solo and small-team builders using Claude Code on agent workflows who want Braintrust-backed evaluation, logging, and trace visibility for assistant behavior.
When should I use braintrustdata-braintrust-claude-plugin?
Use it while shipping and iterating on LLM features—when you need to score responses, capture sessions, and watch regressions after prompt or tool changes.
How do I add braintrustdata-braintrust-claude-plugin to my agent?
Install the plugin marketplace entry from the braintrustdata/braintrust-claude-plugin repository into Claude Code, complete Braintrust account setup, then enable the bundled plugins so hooks and helpers can record and evaluate runs.
Comments
Share how you use braintrustdata-braintrust-claude-plugin, gotchas, or tips for other indie builders.
No comments yet - be the first to share how you use it.