Skip to main content
Buster is a platform for building autonomous AI agents that automate dbt and data engineering workflows. Agents can execute complex end-to-end tasks (like documentation updates, code reviews, schema migrations, data quality monitoring) without human intervention. Agents are defined as YAML configuration files in your repository and run automatically in response to triggers (pull requests, schedules, or custom events). Agent File in Web App

​
Why Buster?

Data engineering with dbt involves countless repetitive tasks that require context about your data, code, and infrastructure:
  • Documentation Drift — dbt models change, but documentation falls behind. Manual updates are tedious and error-prone.
  • Code Review Overhead — PRs need checks for performance issues, breaking changes, naming conventions, test coverage.
  • Schema Changes — Upstream schema changes require updating staging models, propagating to marts, updating tests.
  • Data Quality Monitoring — Regular checks for freshness, null rates, anomalies, referential integrity.
These tasks require understanding both code and data, making them perfect for AI agents with access to your warehouse and repository.

​
Simple Agent Example

Agents are YAML files committed to your repository. Unlike traditional scripts where you write procedural code, agents are declarative—you describe what you want accomplished in natural language. Here’s a complete agent: 
name: docs-updater
description: Automatically updates dbt model documentation when models change

triggers:
  - type: pull_request
    on_changed_files: "models/**/*.sql"

tools:
  preset: standard

prompt: |
  When dbt models change in this PR:
  1. Profile each changed model using the warehouse
  2. Update the YAML documentation with descriptions and stats
  3. Validate the YAML with dbt parse
  4. Commit the changes to this PR branch
Key Components:
  • Triggers — When the agent runs: pull_request, scheduled, event, or manual
  • Tools — What the agent can do: safe (read-only), standard (read/write), or unrestricted (full access including bash)
  • Prompt — Your instructions in natural language describing what to accomplish
When a developer opens a PR that changes a model file, this agent automatically runs, profiles the model, updates documentation, and commits the changes—no manual intervention required.
See the Creating Agents guide for complete configuration options including granular control over permissions, file access patterns, specialized triggers, and more.

​
What’s Next?

Quickstart

Build and deploy your first agent in 10 minutes

How It Works

Learn about architecture, execution, and security

Examples

Explore real agent examples from production users

Installation

Choose between Cloud, CLI, or Self-Hosted deployment