Security audit

Text Compressor

Security checks across malware telemetry and agentic risk

Overview

The main text compression tool is local and mostly purpose-aligned, but the package also includes under-disclosed scripts that archive the whole skill directory.

Install only if you are comfortable with a local text-processing skill that can read input files and write outputs. Review or remove package_manual.py and pack.py before use unless you specifically need local skill packaging, and invoke the compressor only on files you intentionally select.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 88% confidence
Finding: The skill advertises file processing behavior but declares no permissions, despite requiring file read and file write capabilities. This creates a transparency and policy-enforcement gap: the runtime or reviewer may underestimate what the skill can access, increasing the chance of unintended file access or modification.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 94% confidence
Finding: The declared purpose is text compression, but the detected behavior includes packaging the skill, reading and validating SKILL metadata, traversing the full skill directory, and zipping all files. That mismatch is dangerous because hidden archive/traversal functionality can be used to collect and bundle additional files beyond the user's intent, enabling covert data exfiltration or unauthorized repackaging.

Description-Behavior Mismatch

High

Confidence: 90% confidence
Finding: The file's behavior is packaging the entire skill directory into a distributable archive, which does not match the advertised text compression/decompression purpose. This mismatch is dangerous because capability drift and hidden packaging logic can conceal unintended distribution of bundled files, increasing supply-chain and review risk for users who expect only local text processing.

Context-Inappropriate Capability

Medium

Confidence: 86% confidence
Finding: The code creates a zip archive of the whole skill directory, which is broader than necessary for a text-compression skill and can unintentionally bundle sensitive or irrelevant files. In the context of a skill advertised as text cleanup/compression, this unjustified archival capability is more suspicious because users would not reasonably expect redistribution behavior.

Vague Triggers

Medium

Confidence: 81% confidence
Finding: The trigger phrases are broad enough to match common requests like 'optimize' or 'cleanup,' which may cause the skill to activate in situations where the user did not intend file compression or file modification. In combination with file read/write capabilities, overbroad activation increases the risk of unintended invocation and unauthorized processing of local content.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.