Braintrust

Braintrust is a hosted platform for evaluating and tracing LLM applications: it runs a task over a dataset, scores the outputs, stores each run as an experiment and logs production traces. Use when a user asks to write or run a Braintrust eval, score LLM outputs with autoevals, compare prompt or model versions as experiments, run evals in CI and fail on regressions, trace OpenAI or Anthropic calls with Braintrust, manage evaluation datasets, or use the bt CLI.

213 stars 34 forksAI, LLMs & Agents

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add TerminalSkills/skills --skill braintrust
Manual install
git clone --depth 1 https://github.com/TerminalSkills/skills.git /tmp/skills
mkdir -p ~/.claude/skills
cp -r /tmp/skills/skills/braintrust ~/.claude/skills/braintrust

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Braintrust — AI Evaluation and Observability

## Overview

Braintrust measures the quality of LLM applications. An evaluation has three parts — **data** (test cases with `input` and `expected`), a **task** (the function under test) and **scores** (functions that grade each output from 0 to 1). Every run is stored as an experiment that can be compared with earlier runs. The same SDK logs production traces, and the `bt` CLI runs eval files, queries logs and manages datasets from a terminal. The platform is a hosted service (free Starter plan, self-hosting on Enterprise); the SDKs (`braintrust`), the scorer library (`autoevals`) and the CLI are open source. Official SDKs also exist for Go, Ruby, Java and C#; this skill covers TypeScript and Python.

## Instructions

### Install

```bash
npm install braintrust autoevals@latest openai   # TypeScript SDK, scorers, provider client
npm install --save-dev @braintrust/bt tsx        # bt CLI plus a TypeScript runner
pip install braintrust autoevals openai          # Python 3.10+
```

Keep `@latest` on `autoevals`: 0.3.0 declares a pnpm-only `engines` field, so a bare `npm install autoevals` silently resolves to the older 0.0.132 (with `@latest` npm prints an `EBADENGINE` warning and installs 0.3.0). `npx bt --version` prints `bt 0.22.1` or later. `bt` needs a runner for TypeScript files and picks the first it finds of `tsx`, `vite-node`, `ts-node`, `deno`; for Python it uses the active virtualenv.

### Authenticate
Read the full skill on GitHub

Details

Skill name
braintrust
Path
skills/braintrust/SKILL.md
License
Apache-2.0
Repo updated
8 November 2026

FAQ

What does the Braintrust skill do?

Braintrust is a hosted platform for evaluating and tracing LLM applications: it runs a task over a dataset, scores the outputs, stores each run as an experiment and logs production traces. Use when a user asks to write or run a Braintrust eval, score LLM outputs with autoevals, compare prompt or model versions as experiments, run evals in CI and fail on regressions, trace OpenAI or Anthropic calls with Braintrust, manage evaluation datasets, or use the bt CLI.

How do I install the Braintrust skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add TerminalSkills/skills --skill braintrust. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Braintrust popular?

It lives in TerminalSkills/skills, which has 213 GitHub stars and 34 forks.

Related skills

3dsmax Rendering

213

TerminalSkills/skills

Scripts and configures production rendering in Autodesk 3ds Max with the V-Ray and Corona renderers: output size and files, render elements, denoising, light mix, batch and command-line rendering, and network rendering. Use when a user asks to set up a production render, render several cameras in one batch, render from the command line or on a render farm, add render passes for compositing, or cut render time for archviz and product shots.

Other34

3dsmax Scripting

213

TerminalSkills/skills

Covers scripting Autodesk 3ds Max, the 3D modeling and rendering application, with MAXScript and Python (pymxs): scene manipulation, object creation, material assignment, camera and light setup, batch operations, and file I/O. Use when tasks involve automating repetitive 3ds Max workflows, batch processing scenes, running scripts headless with 3dsmaxbatch, creating custom tools, or scripting scene setup for archviz, product visualization, or VFX.

Productivity & Notes34

3proxy

213

TerminalSkills/skills

3proxy is a small open-source proxy server that runs HTTP/HTTPS, SOCKS4/5, SNI and TCP/UDP port-mapping proxies from one config file. Use when a user asks to set up an HTTP or SOCKS5 proxy, add proxy users and passwords, write 3proxy access rules, chain or rotate upstream (parent) proxies, limit bandwidth, connections or monthly traffic per user, run 3proxy in Docker, or fix a 3proxy.cfg that will not start.

DevOps & Cloud34

Browser Use

213

TerminalSkills/skills

Browser Use is an open-source Python library that lets an LLM agent drive a real browser to navigate sites, fill forms, click and extract data from a plain-English task. Use when the user wants an AI agent to automate a website, scrape pages that need clicking or login, return structured data from the web, add browser tools to a coding agent, or run browser agents locally or in the cloud.

AI, LLMs & Agents34

Bolt

213

TerminalSkills/skills

Bolt (bolt.new, by StackBlitz) is an AI app builder that turns a text prompt into a working web or mobile app, with built-in database, hosting and domains. Use when asked to prototype an app from a prompt in Bolt, write better Bolt prompts, plan with Plan Mode, spend fewer tokens, connect GitHub, Supabase, Stripe or Netlify, publish a Bolt project, or run the open-source bolt.diy instead.

AI, LLMs & Agents34

Bug Hunt Swarm

213

TerminalSkills/skills

Finds the cause of a hard bug by running several read-only subagents in parallel, each attacking it from a different angle (change history, code path, data and state, environment, reproduction), then checking their evidence, ranking the hypotheses and naming the one experiment that would confirm the leader. It diagnoses and does not fix. Use when someone says "hunt this bug with multiple agents", "swarm on this regression", "investigate in parallel", "find the root cause", or when a failure is intermittent, crosses components, or has resisted a first look.

AI, LLMs & Agents34

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.