Security audit

Anti-Infinite-Loop Guard

Security checks for vulnerabilities and agentic risk

Overview

This is a low-privilege local loop-guard helper, but its documentation overstates how complete and reliable the protection is.

Install only if you treat this as a simple sample helper, not as a dependable safety control. Review or test the actual anti_infinite_loop.py behavior before relying on it to stop runaway agent work, and consider narrowing the triggers if broad invocation would interrupt normal tasks.

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
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 99% confidence
Finding: The declared description describes a protective control mechanism for preventing infinite loops and repetitive execution cycles. The supplied code does not implement any such behavior. It only sets up a basic CLI, accepts a verbose flag, and prints a static version string. There is no logic for monitoring execution, detecting repetition, tracking progress, enforcing stop conditions, or guarding resources. This is a clear description-behavior mismatch, with the code's primary purpose being a placeholder CLI rather than an anti-infinite-loop guard.
Content: No source excerpt is available for this finding.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 96% confidence
Finding: There is a clear description-behavior mismatch. The declared purpose is a specific safety/control mechanism for preventing infinite loops, but the supplied code is a general-purpose utilities module. While logging and metric helpers could support a loop guard indirectly, this chunk does not itself implement the core declared functionality. Its primary behavior is generic operational support, not anti-loop protection.
Content: No source excerpt is available for this finding.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 98% confidence
Finding: The declared description describes a functional anti-infinite-loop safeguard. However, the provided code chunk contains only placeholder/basic tests with assert True and no operational logic related to loop guarding. Its actual purpose is a trivial test scaffold, which is materially different from the declared skill behavior. There are no undeclared dangerous capabilities, but there is a clear mismatch because the code does not perform the stated function at all.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The trigger phrases are broad, generic operational terms such as 'progress tracking' and 'termination conditions' that could match many normal agent tasks. This can cause unintended invocation of the skill, potentially interrupting valid workflows, forcing premature termination, or altering agent control flow in ways that degrade reliability or availability.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The skill advertises anti-infinite-loop protection but the implementation is only a placeholder CLI that prints a version string and exits. In an agent system, operators may rely on this claimed guard for termination enforcement or resource protection, creating a dangerous mismatch where runaway loops remain unmitigated and can cause resource exhaustion or uncontrolled behavior.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Low

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The docstring and argument parser description present this file as a generic main entry point for 'SKILL_NAME' rather than the declared anti-infinite-loop skill. This is not merely incomplete documentation; it actively identifies the code as something other than the manifest-defined skill.
Content: No source excerpt is available for this finding.