Security audit

日志压缩器

Security checks across malware telemetry and agentic risk

Overview

This skill is for log compression, but it asks for command and write authority over persistent agent memory with broad automatic instructions.

Review this skill before installing. Use it only for explicit log-compression requests, confirm the exact input files, inspect generated summaries before writing them to MEMORY.md, and avoid automatic archive or move operations unless you have backups. Also verify the referenced Node.js script from the publisher before granting exec access, because it is not included in this artifact.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (6)

Intent-Code Divergence

Medium

Confidence: 94% confidence
Finding: The skill presents conflicting claims about network behavior and external dependencies: it says processing is entirely local with no external API use, yet also defines a callback URL, requires an LLM API, and documents API/network failure modes. This can mislead users and host agents into granting the skill more trust or fewer restrictions than appropriate, increasing the risk of unintended data exposure from sensitive logs.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The invocation guidance uses broad natural-language conditions such as using the skill whenever AI models, agent orchestration, or LLM applications are involved. In an agent platform, vague triggers can cause accidental activation during ordinary conversation, leading to unintended file reads, writes, or command execution because this skill has read/exec/write capabilities.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The example trigger phrase '今天的日志太长了，帮我压缩归档' is generic enough to overlap with normal assistant requests. Because the skill is capable of modifying persistent files and running commands, ambiguous triggers increase the chance of unintended autonomous execution on user data.

Vague Triggers

Low

Confidence: 84% confidence
Finding: The batch request phrase '把这周的日志都压缩了' is broad and could trigger multi-file operations without clear scoping or confirmation. In context, this is less severe than arbitrary code execution but still risky because it may cause unexpected processing and subsequent writes or archival actions across many files.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs users to move original logs into an archive directory without a prominent warning that this alters the source data set and may disrupt later workflows or recovery expectations. In agent contexts, destructive or state-changing file operations should be treated as sensitive because the action can be performed automatically at scale on user data.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: Appending generated output directly into MEMORY.md modifies persistent user state without a clear safety warning or review requirement. If the compression output is inaccurate, manipulated, or triggered unintentionally, the agent can permanently contaminate long-term memory and influence future behavior or decisions based on incorrect summaries.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.