Security audit

归档

Security checks for vulnerabilities and agentic risk

Overview

The skill is only a Markdown instruction file, but it mixes unrelated purposes and advertises crawler-evasion capabilities that do not fit a normal Granola archive/search tool.

Install only if you specifically trust the publisher and intend to use it in a tightly controlled, authorized workflow. The main issue is not hidden code, but unclear purpose and advertised evasion-style scraping behavior; users should avoid using it for bypassing access controls, rotating identities, or collecting data without permission.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Behavioral ASTexec() Call, eval() Call, Dynamic Import
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (5)

Intent-Code Divergence

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The skill’s stated purpose is internally inconsistent: it presents itself as a Granola archive/search/sync tool while also claiming SEO optimization and ranking-improvement use cases. This kind of intent ambiguity is dangerous because it can mislead users and downstream agents into applying the skill in contexts beyond its expected scope, reducing the reliability of policy enforcement and risk review.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The documentation advertises anti-crawler bypass, cookie pool management, and IP rotation, which are classic evasion capabilities not necessary for a normal archive/search/sync skill. In context, these features materially increase the risk that the skill could be used to evade access controls, violate site terms, or facilitate stealthy scraping activity.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The skill claims file reading, parsing, and writing support but does not clearly warn users that it may modify local data. In an agent setting with read/exec/glob/grep tools, lack of explicit disclosure can lead to unintended destructive or privacy-impacting operations because users may authorize execution without understanding file-system consequences.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill explicitly advertises command execution but provides no prominent user-facing warning about system-impacting behavior. In this context, command execution is inherently sensitive because it can alter files, processes, network state, or credentials, and the absence of strong guardrails increases the chance of dangerous or abusive use.
Content: No source excerpt is available for this finding.

Ssd 2

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill advertises anti-detection scraping features such as automatic anti-crawler bypass and rotating cookies/IPs, which are strong indicators of evasive operational intent. Even if presented as a premium capability, these features lower the barrier to unauthorized collection and concealment, making misuse substantially more dangerous in an agentic environment.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.