Documentation
Everything you need to understand, evaluate and build BotRampart.
Five guides and one specification. They describe the design as of 18 November 2026: nothing is installable yet, and every page says what is in the first build and what comes later.
Start here
How it works
Budgets, path costs and challenge tiers; why a pass buys credit and never a clearance; the five-minute loop; how decisions are explained.
Install
Getting started
The planned path from npx botrampart init to your first hour of data in observe mode, then to enforcing tier 1.
Guide
Integration guide
SDK options, route costs, sign-in, challenge handling in the browser, the policy file, logs and failure modes, for Node.
Reference
Architecture
Components and licences, the token, sync, hosting on Cloudflare or one container, data layout, the decision model contract, threats and the build plan.
Evaluate
Comparison
Cloudflare, DataDome, Netacea, IntelliFend and open-source peers: sourced, dated, and clear about when to choose them instead.
Specification
OpenAPI spec
The routes the backend SDK mounts on your origin and Core's API: 14 operations with schemas and examples. Every code sample in these docs names its operation.
Three things to know before you read
- BotRampart never labels a client a bot. It meters sessions. If a page seems to be about detection, it is about pricing.
- Nothing sits in front of your site. The SDK runs in your app; your CDN or WAF stays where it is.
- The first build is deliberately small: Node only, challenge tiers 0 to 2, policy in a file, the decision model in shadow mode, and an agent that explains decisions and sends a digest. Pages tag later features later.