# Miasma

> A fast, lightweight server that traps AI web scrapers in an endless loop of poisoned training data to protect your website's content.

Miasma is an open-source Rust-based server designed to fight back against AI web scrapers that harvest website content without permission. It works by serving poisoned training data alongside self-referential links, trapping scrapers in an infinite loop of low-quality "slop" data. The tool is extremely fast and memory-efficient, making it practical to run alongside your existing web infrastructure without significant resource overhead.

- **Poisoned data delivery**: *Miasma proxies content from a configurable poison source and serves it to scrapers, contaminating their training datasets.*
- **Self-referential link trapping**: *Each response includes multiple hidden links pointing back to Miasma, keeping scrapers looping indefinitely.*
- **Hidden link embedding**: *Add invisible HTML links to your site that only scrapers can see, funneling them into Miasma without affecting human visitors or accessibility tools.*
- **Reverse proxy integration**: *Works seamlessly with Nginx and similar reverse proxies — configure a path like `/bots` to route scraper traffic directly to Miasma.*
- **Configurable connection limits**: *Set a maximum number of in-flight requests (`--max-in-flight`) to cap memory usage; excess requests receive a 429 response immediately.*
- **Gzip compression support**: *Optional `--force-gzip` flag compresses all responses regardless of client headers, helping reduce egress bandwidth costs.*
- **CLI configuration**: *All options — port, host, link prefix, link count, poison source URL — are controlled via simple command-line flags.*
- **robots.txt guidance**: *Built-in documentation helps you protect legitimate search engine bots (Googlebot, Bingbot, etc.) from being caught by Miasma.*
- **Install via Cargo**: *Run `cargo install miasma` to get started instantly, or download a pre-built binary from the GitHub releases page.*
- **GPL-3.0 licensed**: *Fully free and open-source; contributions are welcome, though primarily AI-generated pull requests are automatically rejected.*

## Features
- Poisoned AI training data delivery
- Self-referential scraper trapping links
- Hidden HTML link embedding for scraper honeypots
- Reverse proxy integration (Nginx, etc.)
- Configurable max in-flight connections
- Optional gzip compression to reduce egress costs
- CLI-based configuration
- robots.txt guidance for protecting friendly bots
- Minimal memory footprint
- Pre-built binaries and Cargo installation

## Integrations
Nginx, Cargo, robots.txt

## Platforms
LINUX, API, CLI

## Pricing
Open Source

## Version
v0.1.18

## Links
- Website: https://github.com/austin-weeks/miasma
- Documentation: https://github.com/austin-weeks/miasma#readme
- Repository: https://github.com/austin-weeks/miasma
- EveryDev.ai: https://www.everydev.ai/tools/miasma