Local MCP Server

The local MCP server is open source and available on GitHub. If you prefer not to run infrastructure locally, see the Hosted MCP Server instead.

Overview

The Bronto MCP server is a local Python server that exposes Bronto’s dataset and search capabilities to any MCP-compatible AI agent. Search runs over your datasets, which can include trace data. Once running, it gives agents the tools to answer questions against your live telemetry, for example:

“Can you provide some log events from datasets in the ingestion collection, except for the ones related to garbage collection? The data should be between 2025-05-10 16:05:45 and 2025-05-10 16:25:05.”

The server has been tested with Claude Code and should work with any agent that supports the MCP protocol.

Prerequisites

Python 3 installed on your machine
A Bronto API key (see API Keys)
Your Bronto API endpoint URL

Installation

Clone the repository

Check out the project from GitHub:

git clone https://github.com/brontoio/bronto-mcp-server.git
cd bronto-mcp-server

Create a virtual environment

Create a Python virtual environment at the root of the project:

python3 -m venv env

Install dependencies

Install the required packages into the virtual environment:

env/bin/pip install -r requirements.txt

Configuration

The MCP server is configured using environment variables. Set both before starting the server.

Variable	Description	Example
`BRONTO_API_KEY`	A valid Bronto API key	`68ba7ae3-...`
`BRONTO_API_ENDPOINT`	The Bronto API endpoint for your region	`https://api.eu.bronto.io`

Region	API Endpoint
US	`https://api.us.bronto.io`
EU	`https://api.eu.bronto.io`

Starting the Server

Run the following command from the root of the project, substituting your API key and endpoint:

BRONTO_API_KEY=<your-api-key> \
BRONTO_API_ENDPOINT=https://api.eu.bronto.io \
PYTHONPATH=src/main/ \
env/bin/python src/main/brmcpserver/main.py

The server starts on http://localhost:8000 by default.

Keep your terminal session open while using the MCP server. Closing it will stop the server and disconnect any connected agents.

Connecting Claude Code

Once the server is running, register it with Claude Code using the following command:

claude mcp add --transport http bronto http://localhost:8000

This makes the Bronto MCP server available for your current project folder. To make it available globally across all projects, add the --scope user flag:

claude mcp add --transport http bronto http://localhost:8000 --scope user

Full documentation on managing MCP servers in Claude Code is available at docs.anthropic.com.

Verify the connection

Launch Claude Code and run the /mcp command to confirm the Bronto MCP server is available and connected:

/mcp
──────────────────────────────────────
  Manage MCP servers
  1 servers     Local MCPs

  ❯ bronto · ✓ connected

Capabilities

Once connected, the server exposes the following tools to your AI agent.

Tool	Description	When to use
`get_datasets`	Fetches all datasets in your Bronto account, including name, collection, dataset ID, and tags. Tags such as `description`, `service`, `team`, and `environment` are especially useful for understanding dataset contents.	When you need to discover what data sources are available, or find datasets by owner or environment.
`get_datasets_by_name`	Fetches a specific dataset by its exact name and collection name. A dataset is uniquely identified by the combination of both.	When you already know the dataset and collection name and need its log ID or tag metadata.
`get_keys`	Returns all field names present in a specific dataset, identified by its log ID.	To inspect a single dataset’s schema before querying or filtering it.
`get_all_datasets_keys`	Returns a map of dataset IDs to their field names across all datasets.	To identify which datasets contain a particular field, or to determine the exact field name from a description.
`get_key_values`	Returns the available values for a specific field in a specific dataset.	To discover valid field values and build precise filters before running a search.
`search_logs`	Searches raw events across one or more datasets over a time range. Supports SQL-style `WHERE` filters, multiple dataset IDs, and returns full events including the `@raw` field.	Event-level investigations where you need to inspect individual entries — for example, finding errors or warnings in a specific service.

Frequently Asked Questions

Which AI agents are supported? The server implements the MCP protocol and should work with any MCP-compatible agent. It has been tested with Claude Code. Can I make the server available across all my projects? Yes. Add --scope user when registering the server with Claude Code and it will be available globally rather than only for the current project folder. Where do I find my API key? API keys can be created and managed in Settings → API Keys in the Bronto UI. See API Keys for details. What is the difference between the local server and the hosted MCP server? The local server runs on your machine and requires Python, a virtual environment, and a running terminal session. The hosted MCP server is managed by Bronto — there is nothing to install or run locally. For most teams the hosted option is simpler; the local server is useful if you need to run in an air-gapped environment or want to inspect or modify the server code directly.

Getting Started

MCP

AI Investigations

LLM Observability

LLM-Powered Parsing

AI Integrations

Overview

Prerequisites

Installation

Configuration

Starting the Server

Connecting Claude Code

Verify the connection

Capabilities

Frequently Asked Questions

​Overview

​Prerequisites

​Installation

​Configuration

​Starting the Server

​Connecting Claude Code

​Verify the connection

​Capabilities

​Frequently Asked Questions

Overview

Prerequisites

Installation

Configuration

Starting the Server

Connecting Claude Code

Verify the connection

Capabilities

Frequently Asked Questions