Security audit

Telegram身体扫描工具

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a disclosed body-scan workflow, but it handles very sensitive body-video data while requesting broad tool authority and leaving privacy/storage boundaries unclear.

Review before installing. Only use this with adults who explicitly consent, avoid uploading private or local files, and require a clear AnthroVision bridge privacy policy covering what videos and measurements are sent, where they are stored, retention/deletion, and who can access them. The publisher should also narrow the declared read/write/exec authority and remove unrelated security-scanning claims.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The skill is presented as a Telegram body-scan workflow, but the manifest advertises unrelated capabilities such as CVE correlation, threat intelligence, compliance auditing, and asset risk scoring. This kind of scope inflation is dangerous because it can justify broader tool use or data handling than users expect, increasing the risk of unauthorized security actions, misleading trust assumptions, or privilege misuse in an agent environment.

Description-Behavior Mismatch

Medium

Confidence: 90% confidence
Finding: The documentation claims generic file processing, API integration, and system command execution even though the described workflow is limited to submitting videos and polling scan status. In an agent setting, broad statements like these can normalize unnecessary read/write/exec behavior and increase the chance that the skill is granted or uses dangerous capabilities beyond its legitimate purpose.

Intent-Code Divergence

Medium

Confidence: 82% confidence
Finding: The FAQ says storage policy depends on bridge configuration, while the security declaration later implies encrypted storage and stronger controls. This contradiction is risky for a body-video processing skill because users may rely on privacy guarantees that are not actually defined or enforced, leading to mishandling of sensitive biometric and body-measurement data.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The description discusses consent and processing flow but does not clearly warn that users' body videos and resulting measurements are transmitted to an external bridge/service. Because this is highly sensitive personal data, omitting an explicit transfer notice undermines informed consent and may expose users to privacy, compliance, and trust risks.

Static analysis

No suspicious patterns detected.