Security audit

Adaptive Reasoning

Security checks for vulnerabilities and agentic risk

Overview

This skill broadly changes how the agent decides when to think harder, but its behavior is disclosed, non-executing, and aligned with that purpose.

Install this only if you want the agent to automatically evaluate every request for reasoning depth. It should not access data or run code, but it may change response style, token use, and final formatting unless the user disables or overrides it.

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 (3)

Vague Triggers

High

Confidence: 98% confidence
Finding: The skill declares that it triggers on every user message, making it a global pre-processing hook rather than a narrowly scoped helper. Broad trigger scope increases the chance of unintended behavior, instruction interference, and unnecessary policy/state changes across unrelated conversations, especially when paired with automatic mode switching.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The automatic activation section instructs the agent to enable reasoning silently and 'do not ask,' but it does not define robust safety boundaries or precedence rules. This can cause unauthorized state changes, inconsistent behavior across tasks, and conflicts with user preferences or higher-priority system controls.

Natural-Language Policy Violations

Medium

Confidence: 92% confidence
Finding: The skill mandates appending visible reasoning-state icons to responses without user opt-in. While not a direct code-execution risk, it creates unsolicited output modification, can leak internal mode/state decisions, and may interfere with downstream formatting, automation, or product UX expectations.

Static analysis

No suspicious patterns detected.