Security audit

MOA-Debate

Security checks for vulnerabilities and agentic risk

Overview

The skill is a transparent debate-preparation workflow that uses bounded model calls and does not request local files, credentials, persistence, or system access.

Install this if you want a structured, multi-round debate simulator. Be aware that broad requests like argument stress-testing may trigger the full workflow and consume more model calls than a short critique, so specify a low round cap or ask for lighter analysis when needed.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The skill activates on broad phrases like 'stress-test an argument' or 'prepare for a formal debate', which overlap with many ordinary persuasion, writing, and analysis requests. This can cause the debate orchestration to trigger unexpectedly, leading the agent to make multiple sequential model calls and adopt adversarial roles when the user did not clearly request that workflow.

Natural-Language Policy Violations

Low

Confidence: 89% confidence
Finding: The prompt says 'Return ONLY the question,' and the surrounding examples and all embedded agent instructions are written as fixed English-language output templates. For a general-purpose debate skill, this effectively constrains responses to English without any documented user opt-in or locale choice, which can violate language/locale policy expectations.

Static analysis

Detected: suspicious.prompt_injection_instructions

Prompt-injection style instruction pattern detected.

Warn

Code: suspicious.prompt_injection_instructions
Location: references/DEBATE_FORMAT.md:38

Prompt-injection style instruction pattern detected.

Warn

Code: suspicious.prompt_injection_instructions
Location: SKILL.md:41