Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Verifieddanny/BurnGuard/llms.txt

Use this file to discover all available pages before exploring further.

BurnGuard sits quietly between your application and every AI provider — intercepting each API call, counting tokens as they stream through, calculating cost in real time, and blocking requests the instant your budget is exceeded. No 8-hour delays. No $6K/year FinOps tools. One binary.

Quickstart

Install BurnGuard and protect your first AI application in under five minutes.

How It Works

Understand the proxy architecture, token counting, and budget enforcement flow.

Configuration

Configure providers, budget limits, alerts, and cloud sync via burnguard.yaml.

Cloud Dashboard

Sign up at burnguard.run for spend charts, model breakdowns, and request logs.

​
Get running in minutes

1

Install the binary

Download BurnGuard for your platform — macOS, Linux, or Windows. No runtime required.
# macOS (Homebrew)
brew tap verifieddanny/tap
brew install burnguard

# macOS / Linux
curl -sSL https://burnguard.run/install.sh | sh

# Windows (PowerShell)
irm https://burnguard.run/install.ps1 | iex
2

Run the setup wizard

burnguard init walks you through selecting providers, setting a budget, and optionally connecting your cloud dashboard — no YAML editing required.
burnguard init
3

Start the proxy

The proxy listens on localhost:8080 and forwards requests to your configured AI providers.
burnguard start
4

Point your app at BurnGuard

Change one line in your application — the base URL. Every request now flows through BurnGuard.
# Anthropic
client = Anthropic(base_url="http://localhost:8080/anthropic")

# OpenAI
client = OpenAI(base_url="http://localhost:8080/openai/v1")

​
Key features

Real-Time Token Counting

Tokens counted as they stream — including SSE streaming responses — not hours later.

Hard Budget Caps

Requests blocked instantly with HTTP 403 when your limit is hit, before they reach the provider.

Multi-Provider Support

Route to Anthropic Claude and OpenAI GPT via path prefix — /anthropic/... and /openai/....

Slack & Discord Alerts

Webhook notifications at 50%, 80%, and 100% of budget — configurable thresholds.

Persistent Memory

SQLite stores usage locally so budget enforcement survives proxy restarts.

Cloud Dashboard

Sync to burnguard.run every 60 seconds for team-visible charts and request logs.

​
Explore the docs

Integrate Anthropic

Step-by-step guide for routing Claude requests through BurnGuard.

Integrate OpenAI

Route GPT requests through BurnGuard with one base URL change.

CLI Reference

Full reference for burnguard init and burnguard start.

API Reference

REST API used by the dashboard and proxy sync.

Pricing Tables

Built-in per-model cost rates for Anthropic and OpenAI.

Self-Hosting

Run the BurnGuard backend API and dashboard on your own infrastructure.

Build docs developers (and LLMs) love

Get started for freeTalk to us