Security audit

ArifOS Sense

Security checks for vulnerabilities and agentic risk

Overview

This skill is a governance tool, but it gives itself broad control over the agent and creates persistent cross-session decision logs with unclear user control.

Install only if this is specifically for Arif's own arifOS workflow and you want a strict governance layer that can pause actions and keep persistent audit logs. Other users should review or modify it first so authority belongs to the authenticated user, activation is explicit, and any memory logging is opt-in, scoped, redactable, and deletable.

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

Findings (2)

T01 · Skill Instruction Hijacking

Error

Location: SKILL.md:38
Finding: Non-Negotiable Governance Instructions Hijack Agent Control and Approval Authority
Content: View full analysis
Remediation: View remediation

T02 · Agent Memory Poisoning

Error

Location: SKILL.md:47
Finding: Persistent Governance Ledger Can Poison Agent State Across Sessions
Content: View full analysis
Remediation: View remediation

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

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The trigger list in metadata is extremely broad and includes common words like "evaluate," "hold," and "is this safe," which can cause the governance skill to activate in many unrelated conversations. In an agent system, this kind of over-triggering can unpredictably override other skills, block normal actions, or inject governance behavior into benign tasks, creating denial-of-service and control-flow manipulation risk.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 91% confidence
Finding: The "When to Trigger This Skill" section uses subjective conditions such as requests involving external systems, irreversible actions, or the appearance of words like "evaluate" and "HOLD" without precise scoping. Ambiguous activation criteria make the skill hard to predict and can let adversarial prompts intentionally invoke or suppress the governance layer, weakening reliability and potentially interfering with other safety or task-routing mechanisms.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The file includes Indonesian terms such as "Amanah," "Hidayat," and the mixed-language phrase "From weakest to strongest约束" as core operating vocabulary, but it does not state that the user opted into this language or provide an alternative. This can violate a language/locale policy when a skill imposes terminology the user may not understand.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.