Guides

Configuration you can copy, with the reasoning behind it and an honest note on what each measure does not cover.

AI crawler policy

Block AI training crawlers without losing search traffic

GPTBot and ClaudeBot crawl for training. OAI-SearchBot and PerplexityBot decide whether assistants can cite you. Blocking all of them costs referral traffic — here is how to separate the two, with the exact directives.

Reference

The complete robots.txt list for AI crawlers

A named-crawler reference: which user agents crawl for training, which power AI search citations, which ignore robots.txt entirely, and the exact directives that work for each.

Browser boundary

The security headers checklist: CSP, HSTS and the five that actually matter

Most guides list twenty directives you will never ship. These are the ones that block real attacks, what each breaks when misconfigured, and how to roll them out without downtime.

Abuse boundary

Rate limiting patterns that survive real traffic

Fixed windows let attackers double their burst at the boundary. Token buckets need shared state. How to pick a limiter that holds up on serverless, at the edge, and behind a CDN.

Architecture

WAF vs application middleware: where should your security boundary live?

Edge WAFs catch traffic before it reaches you but cannot see application context. Middleware sees everything but runs on your compute. The honest tradeoff, and when running both is worth it.

AI boundary

Prompt injection defense for web applications that call LLMs

If your app passes user input, page content or tool output into a model, you have an injection surface WAF rules do not cover. What actually helps, and what only looks like it helps.