---
id: RES-001
title: Arcanum-Sec Resources
slug: arcanum-sec-resources
type: lesson
code: RES-001
status: published
category: resources
pillar: resources
section: references
collection: arcanum-sec
tags: [arcanum, resources, taxonomy, references]
---

# Arcanum-Sec Resources

Bot-Tricks lesson content is shaped by Arcanum-Sec's AI security research
and our own lab discoveries. This page lists the primary resources that
inform our curriculum.

## AI Security Resources

[https://arcanum-sec.github.io/ai-sec-resources/](https://arcanum-sec.github.io/ai-sec-resources/)

A comprehensive collection of AI security tools, techniques, and research.
This resource covers the full landscape of AI security — from prompt
injection to model theft, from data poisoning to supply chain attacks.

We use this as a primary source for identifying new threat vectors and
shaping lessons around real-world findings.

## Prompt Injection Taxonomy

[https://arcanum-sec.github.io/arc_pi_taxonomy/](https://arcanum-sec.github.io/arc_pi_taxonomy/)

A structured taxonomy of prompt injection techniques. This is the backbone
of our lesson organization. The taxonomy maps techniques to:

- Attack surfaces (direct, indirect, tool-mediated)
- Evasion methods (encoding, obfuscation, context manipulation)
- Defense categories (filtering, isolation, monitoring)

Our Techniques, Evasions, and Fundamentals families map directly to
categories in this taxonomy.

## How We Use These Resources

- **Lesson topics** are selected from the Arcanum PI Taxonomy to ensure
  comprehensive coverage of the prompt injection landscape.
- **Techniques and evasions** are mapped to real-world findings documented
  in Arcanum research, with citations where applicable.
- **Lab scenarios** are built from real CVEs and vulnerabilities documented
  in Arcanum research — each Arcanum IRL lab simulates a real-world app
  with a specific, documented vulnerability class.
- **New content** is added as Arcanum publishes new findings, keeping our
  curriculum current with the evolving AI security landscape.

## Our Own Lab Discoveries

In addition to Arcanum research, our own lab development surfaces new
techniques and evasion patterns. These discoveries feed back into the
lesson catalog as standalone lessons or as additions to existing technique
families.

Examples:
- Our Instaglam lab revealed how side-channel inboxes can be exploited
  through indirect prompt injection
- Our Chevrolite lab demonstrated persona adoption as a multi-step
  exploit chain
- Our BingBong lab showed how encoding bypasses surface-level filters

## Partner

Learn more about Arcanum Information Security at
[arcanum-sec.com](https://arcanum-sec.com).

Featured in Jason Haddix's Attacking AI Course.