ClawHub
Back to skill

Security audit

geo-optimization-tool

Security checks for vulnerabilities and agentic risk

Overview

This skill is a GEO analysis helper with a small API-query script; it has some scope and documentation caveats but no evidence of hidden, persistent, destructive, or credential-seeking behavior.

Install only if you want a GEO-focused assistant with a China-oriented emphasis and are comfortable with optional read-only calls to the listed external API. Avoid sending confidential brand plans, private drafts, customer data, or secrets to the documented API endpoints unless the publisher provides clearer privacy and data-flow terms.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Vague Triggers

High
Confidence
95% confidence
Finding
The trigger list contains very broad phrases such as common AI- or SEO-related terms, making accidental invocation likely in unrelated conversations. Overbroad auto-invocation is dangerous because it can cause the skill to activate, steer outputs, or surface external-script/API instructions without meaningful user intent or contextual fit.

Lp3

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding
The skill advertises an executable shell script but does not declare any tool scope or permissions, creating a mismatch between documented capabilities and the security boundary visible to users and the platform. This can lead to unexpected shell access or unauthorized execution paths if the runtime infers or permits shell use despite the missing declaration.

Natural-Language Policy Violations

Medium
Confidence
89% confidence
Finding
The title and opening instruction position the skill as helping "Chinese brands" and specializing in Chinese AI engines, which imposes a locale-specific operating context by default. There is no language or locale choice offered to the user, nor an explicit opt-in mechanism for this regional focus.

Intent-Code Divergence

Medium
Confidence
93% confidence
Finding
The API backend is presented as a GEO engine database, but the listed endpoints include unrelated compliance and replacement-suggestion functions, which is a direct contradiction in stated purpose. Such contradictions are a red flag because they can conceal secondary processing logic or a multi-purpose backend that handles sensitive text in ways the user did not consent to.

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The documented API functions include compliance-checking and suggestion endpoints that do not align with the stated GEO monitoring purpose, indicating the skill may invoke an API with broader or different behavior than users expect. Security-relevant capability drift is dangerous because hidden or mislabeled backend functionality can exfiltrate data, process unexpected content, or expand the attack surface under a benign-looking skill description.

Static analysis

No suspicious patterns detected.