Security audit

Make Friends

Security checks for vulnerabilities and agentic risk

Overview

This is a local friendship-planning skill that stores social notes for continuity but shows no code execution, credential access, network use, or deceptive behavior.

Before installing, understand that the skill may keep local notes about your social goals, habits, promising contacts, and follow-up plans in ~/make-friends/. Avoid storing sensitive details about other people, and use the paused or never_ask status if you do not want proactive social reminders.

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

Medium

Confidence: 95% confidence
Finding: The activation rules are broad enough to trigger on common emotional or lifestyle cues like loneliness, moving, or hobbies, which can cause the skill to engage without clear user intent. In a skill that also persists durable notes, overbroad activation increases the chance of collecting sensitive social-context data unexpectedly and can blur consent boundaries.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The setup instructs the agent to store durable personal context for future sessions without requiring an upfront disclosure before memory is created. Because the stored data includes social energy, habits, promising contacts, and recurring blockers, this can create non-obvious profiling and privacy risks if users do not realize persistence is occurring.

Vague Triggers

Low

Confidence: 88% confidence
Finding: This markdown file defines a memory template but provides no explicit trigger phrases, activation conditions, or exclusion criteria for when the skill should apply. That ambiguity can cause unintended invocation because auditors and integrators cannot tell when the template is meant to be used versus when it should not.