SecuraAI · Project Feral

Join Phase II Research

We're building a community research initiative around agentic AI security. Phase I identified 10 architectural threats across 7 MAESTRO layers. Phase II goes hands-on — red teaming, scanning, and defensive tooling against a controlled OpenClaw fork.

—

Researchers Joined

Critical

High

Medium

10/10

ASI Coverage

7/7

MAESTRO

Sign Up

Phase II

Name *

Email *

Role (optional)

Organization (optional)

GitHub (optional — helps us review contributor applications)

I'd like to contribute to Phase II research — not just receive updates, but participate as a researcher in the initiative.

Contributor Details

Research Track

Experience

Brief Background (optional)

We'll notify you when Phase II launches. Contributors receive a separate onboarding invitation.

✓

You're In

Thanks for joining Project Feral. We'll be in touch when Phase II launches. In the meantime, explore the Phase I findings in the assessment portal.

Phase II Research Tracks

⚡

Red Team Testing

Prompt injection, tool abuse, privilege escalation, and multi-agent attack chains against a sandboxed fork.

ACTIVE TESTING

⊞

Vulnerability Scanning

SAST, DAST, dependency scanning, and agentic-specific methodologies across the codebase.

AUTO + MANUAL

◎

Defensive Tooling

Detection rules, sandboxing prototypes, and monitoring approaches for agentic AI systems.

OPEN SOURCE

◇

Comparative Analysis

How threats differ across OpenClaw derivatives and architectural variants.

RESEARCH