Security audit

Compact Guard

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed context-saving skill, but users should review what it writes to memory before using it in sensitive sessions.

Install this only if you want the agent to save selected conversation and tool-output details into local memory before compaction. Avoid using it in sessions containing secrets, tokens, customer data, or confidential internal details unless you review and redact what will be stored.

SkillSpector

Vulnerability Patterns

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

Findings (2)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger conditions are broad enough to allow unsolicited or automatic activation when context is merely 'close to full' or when a heartbeat heuristic fires. In this skill, activation leads to scanning prior tool outputs and writing extracted content into persistent memory files, so overbroad triggers increase the chance of unintended data persistence without clear user consent.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill instructs the agent to extract facts, decisions, errors, preferences, and progress from conversation/tool output and append them into dated memory files, but the description does not prominently warn users that persistent storage will occur. In context, this is more dangerous because the retained material may include sensitive paths, endpoints, configuration values, or user preferences that were safe in transient context but risky when durably stored.

Static analysis

No suspicious patterns detected.