Security audit

Industry News Summary

Security checks for vulnerabilities and agentic risk

Overview

This skill fetches public AI news and writes a local dated summary file, with no evidence of credential access, exfiltration, destructive actions, or hidden unrelated behavior.

Before installing, be aware that running the skill may leave daily AI news summaries in a local memory/ folder. This is ordinary local caching of public web-source summaries, but users with strict retention, audit, or workspace-cleanup requirements should delete those files or avoid the skill until caching is better documented or optional.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 94% confidence
Finding: The skill’s documented purpose is to fetch and summarize current AI news, but the implementation notes introduce persistent local storage of generated summaries in memory/ without clearly declaring that behavior in the primary description. Hidden or under-disclosed persistence changes the trust and data-handling model of the skill, creating risk of unintended retention, later disclosure, or misuse of stored content.

Description-Behavior Mismatch

Medium

Confidence: 91% confidence
Finding: The documentation suggests the skill only performs a web_search-based summary, but the implementation notes add persistent caching to local memory files. This discrepancy can mislead users and reviewers about the skill’s effective capabilities, reducing informed consent and making unintended data retention more likely.

Context-Inappropriate Capability

Medium

Confidence: 88% confidence
Finding: Local memory/file caching is not necessary for the stated task of generating a daily news summary and therefore expands the skill’s capabilities beyond least privilege. Even if the content is public news, unnecessary persistence increases attack surface and can create accumulation of files, metadata leakage, or unexpected reuse in future agent workflows.

Missing User Warnings

Low

Confidence: 83% confidence
Finding: The skill omits a user warning that it writes cached summaries to local storage, which undermines transparency about side effects. While the data here is likely low sensitivity, undisclosed writes can still violate user expectations and become more dangerous when skills are reused in broader environments with shared memory or audit requirements.

Static analysis

No suspicious patterns detected.