Security audit

Gv Caller

Security checks for vulnerabilities and agentic risk

Overview

This skill openly automates Google Voice calls, but it uses stored Google session cookies and broad triggers without clear per-call confirmation controls.

Review before installing. Use only with a dedicated low-risk Google Voice account in an isolated environment, disable broad auto-invocation, protect or avoid reusable Google cookies, and require explicit human confirmation of the exact phone number, message or audio, duration, and expected cost before every call.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The skill reads persistent Google Voice session cookies from a fixed filesystem path and injects them into a browser session, effectively reusing stored authentication material without any visible consent, scoping, or secret-management controls. In the context of a tool that places outbound calls, this enables account misuse and unauthorized actions if the cookies belong to another user or are unexpectedly present in the shared workspace.

Vague Triggers

High

Confidence: 97% confidence
Finding: The invocation patterns are broad natural-language phrases like '打电话给', '给.*打电话', '拨通', and 'call', which are likely to match ordinary conversation and trigger a high-risk capability unintentionally. In this skill's context, accidental activation is especially dangerous because the action initiates real outbound phone calls using stored Google Voice session cookies and can transmit synthesized or local audio to external parties.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: Accessing authentication cookies from a sensitive file without disclosure is dangerous because cookies are bearer credentials that can grant full account access for telephony actions. In this skill's context, undisclosed credential reuse is especially risky because it can silently place calls under an existing Google Voice account, leading to privacy, billing, and abuse consequences.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The code navigates to Google Voice and initiates an outbound call automatically with the provided number, creating an external network action with real-world effects and no explicit confirmation step. Given the skill's purpose is automated calling, the absence of a warning or approval gate materially increases the risk of accidental, unauthorized, or abusive calls.

Missing User Warnings

Low

Confidence: 84% confidence
Finding: Saving a screenshot of the call state to disk may capture sensitive information such as phone numbers, account details, recent contacts, or call metadata, and it is done without disclosure or retention controls. While lower severity than credential misuse, it still creates a privacy and data-handling risk because artifacts may persist in shared or insecure storage.

Static analysis

No suspicious patterns detected.