SERPENT

Chain-of-Thought Attack Testing

Reasoning is the new attack surface. SERPENT finds the flaws. CoT inflation, steganographic detection, reasoning chain manipulation, data leakage via reasoning, reasoning loop injection, and full CoT audit — weaponised for authorised red team engagements.

6
Subsystems
61
Tests
View Documentation GitHub

Reasoning Is the New Attack Surface

SERPENT targets chain-of-thought reasoning — the internal deliberation that modern AI models use to solve complex problems. Every reasoning step, every intermediate conclusion, every thought chain — all exploitable. SERPENT finds the flaws in how AI thinks.

01

INFLATOR

COT INFLATION

Inflate chain-of-thought reasoning to consume excessive tokens. Resource exhaustion via reasoning amplification. Cost manipulation through deliberate reasoning expansion.

02

STEGO

STEGANOGRAPHIC DETECTION

Detect hidden information encoded within reasoning chains. Steganographic payload identification. Covert channel detection in chain-of-thought outputs.

03

HIJACKER

REASONING CHAIN MANIPULATION

Manipulate intermediate reasoning steps to alter final outputs. Inject false premises into reasoning chains. Redirect logical conclusions through crafted inputs.

04

EXFILTRATOR

DATA LEAK VIA REASONING

Extract sensitive data through reasoning chain outputs. System prompt leakage via reasoning traces. Training data extraction through deliberate reasoning exploitation.

05

LOOPER

REASONING LOOP INJECTION

Inject infinite reasoning loops. Denial-of-service via circular logic. Resource exhaustion through recursive reasoning patterns. Self-referential chain exploitation.

06

AUDITOR

FULL COT AUDIT

Complete chain-of-thought security audit. Reasoning integrity verification. Thought chain consistency checking. Full CoT vulnerability assessment and reporting.

UNLEASHED Gate

Standard mode detects. UNLEASHED exploits. Ed25519 crypto. Dual-gate safety. One operator.

Detection

Maps chain-of-thought attack surfaces. Identifies vulnerable reasoning patterns. No exploitation. Reports only.

Dry Run

Plans full CoT exploitation campaigns. Shows exactly what would work. Ed25519 required. No execution.

Live Execution

Cryptographic override. Private key controlled. One operator. Founder's machine only.

THIS TOOL IS FOR AUTHORISED SECURITY TESTING ONLY. EVERY EXECUTION IS SIGNED AND LOGGED.

61
Tests
6
Subsystems
50,914
Ecosystem Tests
Available On

Security Distros & Package Managers

Kali Linux
.deb package
Parrot OS
.deb package
BlackArch
PKGBUILD
REMnux
.deb package
Tsurugi
.deb package
PyPI
pip install

Reasoning Is the New Attack Surface. SERPENT Finds the Flaws.

6 subsystems. 61 tests. Chain-of-thought exploitation. The tool that proves your AI reasoning isn't safe.