Security audit

Call Geo Agent

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed marketing/GEO workflow that uses external model/search tools and project documents, with privacy and data-retention cautions but no evidence of hidden or destructive behavior.

Install only if you are comfortable sending the relevant brand or product materials to external model/search tools and having prompts, responses, URLs, and generated content saved in project documents. Avoid using confidential launch plans, trade secrets, credentials, private customer data, or unreleased strategy unless you have approval to share and retain that information.

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

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (5)

Missing User Warnings

High

Confidence: 98% confidence
Finding: The workflow explicitly requires transmitting user-provided brand/product materials, generated keyword lists, and prompt variants to online/model tools, including `call_gpt_5_online`, while prohibiting modification or filtering of returned content. Without warning or minimization, this creates a direct risk of exposing confidential product strategy, marketing plans, or proprietary inputs to external services.

Natural-Language Policy Violations

Medium

Confidence: 83% confidence
Finding: The skill description and workflow are written as mandatory Chinese-language instructions and frame the agent role entirely in Chinese, with no indication that users may choose another language. This creates a language policy concern because the skill appears to enforce a specific locale without opt-in or justification.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The workflow explicitly requires transmitting user-provided brand/product materials, generated keyword lists, and prompt variants to online/model tools, including `call_gpt_5_online`, while prohibiting modification or filtering of returned content. Without warning or minimization, this creates a direct risk of exposing confidential product strategy, marketing plans, or proprietary inputs to external services.

Natural-Language Policy Violations

Medium

Confidence: 83% confidence
Finding: The main instruction block is written as mandatory Chinese-language operational guidance and does not offer any language-selection mechanism. Without user opt-in or a documented regional justification, this is a natural-language locale policy issue.

Ssd 3

Medium

Confidence: 97% confidence
Finding: The instructions require comprehensive project logging and verbatim recording of all prompts, model responses, and URLs, while also forbidding filtering or omission. This creates unnecessary retention of potentially sensitive user inputs, proprietary research, and third-party references in persistent project documents, increasing exposure if documents are shared, indexed, or submitted.

Static analysis

No suspicious patterns detected.