How to Use the Braintrust MCP in Mastra AI

Q: How do I register Braintrust tools in a Mastra AI agent?

Install @mastra/mcp@latest and configure the client with your Vinkius HTTP URL. Call mcpClient.listTools() to get the tool definitions, then spread them directly into your agent's tool array.

Q: Can I require manual approval before Mastra AI modifies my Braintrust datasets?

Yes, you can set requireToolApproval on your Mastra agent configuration. This stops the agent from running insertdatasetrow or createproject until a human reviews and approves the payload.

Q: How do I find my existing evaluation projects?

Your agent can run listprojects to discover all active evaluation environments. This allows the workflow to dynamically select the correct project ID before starting a test run.

Q: How is my Braintrust API key secured?

All credentials used by listenvvars are encrypted at rest and never exposed to the agent's log output. The Vinkius runtime executes each tool call in a zero-trust, ephemeral environment that destroys all memory traces immediately after the workflow step completes.

Build resilient, auto-retrying evaluation workflows in Mastra AI using this Braintrust MCP Server.

See Vinkius in Action

Works with every AI agent you already use

…and any MCP-compatible client

MCP Servers - Free for Subscribers

Connect Braintrust MCP to Mastra AI

Create your Vinkius account to connect Braintrust to Mastra AI and route execution through our secure gateway. The platform manages server hosting, runtime updates, and security layers. Configuration requires no manual server provisioning.

GDPR Free for Subscribers

Setup Braintrust with Mastra AI

Ask AI about this MCP

ChatGPT

Claude

Perplexity

Automate test pipelines with Mastra AI workflows

This Braintrust MCP Server enables your workflows to spin up new evaluation runs using the `create_experiment` tool. By combining Mastra workflows with these actions, you can run evaluations that automatically retry with exponential backoff if a model rate-limit hits. This workflow engine handles conditional branching natively. If an evaluation score drops below your threshold, you can trigger a branch to notify your engineering team immediately.

Manage prompts inside agent workflows

The Braintrust MCP Server provides direct access to your version-controlled instructions through the `get_prompt` and `list_prompts` tools. Your Mastra agents can pull the exact instructions they need for a specific run dynamically. This setup prevents prompt drift across your systems. You update the template in one place, and your autonomous agents pick it up on their next run.

Dynamically update test datasets

This Braintrust MCP Server allows your automated pipelines to update test cases using the `insert_dataset_row` tool. Your agents can append new test cases to an existing matrix based on real-world failures. This creates a closed-loop system where your production application continuously feeds its own evaluation suite. You spend less time manually writing test cases.

Setup guide

Set up Braintrust MCP in Mastra AI

Prerequisites

Node.js 18+ and a TypeScript project
@mastra/mcp + @mastra/core packages
Active Vinkius subscription with a valid endpoint token

1

Install dependencies
Run npm install @mastra/mcp @mastra/core plus your preferred model provider (e.g. @ai-sdk/openai).
2

Configure the MCPClient
Create an MCPClient with your Vinkius endpoint as a URL object. Replace [YOUR_TOKEN_HERE] with your token from cloud.vinkius.com.
3

Discover and inject tools
Call mcpClient.listTools() and spread the result into your agent's tools object. All Braintrust tools become native Mastra tools.
4

Run with any model
Swap openai("gpt-4o") for any AI SDK-compatible provider. Call agent.generate() and the agent routes tool calls through MCP automatically.

agent.ts

import { MCPClient } from "@mastra/mcp";
import { Agent } from "@mastra/core/agent";
import { openai } from "@ai-sdk/openai";

const mcpClient = new MCPClient({
  id: "braintrust-mcp-client",
  servers: {
    "braintrust-mcp": {
      url: new URL(
        "https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp"
      ),
    },
  },
});

const agent = new Agent({
  name: "Braintrust Agent",
  model: openai("gpt-4o"),
  instructions: "You have access to Braintrust tools.",
  tools: {
    ...(await mcpClient.listTools()),
  },
});

const result = await agent.generate(
  "List recent Braintrust transactions"
);
console.log(result.text);

Get your connection token →

Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Braintrust. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.

Why Choose Vinkius

Vinkius connects your tools to AI with real-time monitoring and automatic cost savings — all from one dashboard.

Connect Braintrust now

Real-time monitoring

Live

visibility into every interaction

Connect your favorite tools to your AI and see exactly what's happening — every request, every response, in real time.

Built-in savings

60%

lower AI costs

Vinkius compresses data between your apps and your AI automatically. Lower bills every month — no configuration required.

Single dashboard

One

place for every integration

Every tool your AI connects to, managed from a single screen. One account, complete control.

Common questions about Braintrust MCP in Mastra AI

Install `@mastra/mcp@latest` and configure the client with your Vinkius HTTP URL. Call `mcpClient.listTools()` to get the tool definitions, then spread them directly into your agent's tool array.

Yes, you can set `requireToolApproval` on your Mastra agent configuration. This stops the agent from running `insert_dataset_row` or `create_project` until a human reviews and approves the payload.

The client automatically detects the transport layer. It supports both standard Server-Sent Events (SSE) and Streamable HTTP, allowing your workflows to communicate with the MCP Server without manual socket configuration.

Your agent can run `list_projects` to discover all active evaluation environments. This allows the workflow to dynamically select the correct project ID before starting a test run.

All credentials used by `list_env_vars` are encrypted at rest and never exposed to the agent's log output. The Vinkius runtime executes each tool call in a zero-trust, ephemeral environment that destroys all memory traces immediately after the workflow step completes.

Use it with your favorite AI tools

Connect this server to Cursor, Claude, VS Code, and more.

OpenAI Agents SDK sdk-python

Google ADK sdk-python

Pydantic AI sdk-python

Vercel AI SDK sdk-typescript