ClawHub
Back to skill

Security audit

让飞书可以发图片消息

Security checks across malware telemetry and agentic risk

Overview

This skill appears to be a straightforward Feishu image-message sender, with no hidden persistence or unrelated behavior found, but its documentation overstates some supported features.

Install only if you are comfortable giving these scripts a Feishu app ID and secret that can send messages under that app's permissions. Use a least-privilege Feishu app, keep the secret out of repositories and logs, use trusted image URLs, and be aware that chat_id/user_id delivery may not work until receive_id_type is made configurable.

SkillSpector

By NVIDIA
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 (3)

Tp4

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding
The skill description overstates functionality and understates actual behavior, which can mislead users about what data flows occur and what identifiers are supported. Security-relevant mismatches reduce user ability to assess risk and can cause messages or data to be sent in unintended ways, especially if the implementation hardcodes receive_id_type=open_id while claiming broader recipient support.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
Fetching remote image URLs means the skill may contact arbitrary third-party hosts, exposing the user's IP/network metadata and pulling untrusted content into the workflow. Sending the fetched image and accompanying text to Feishu is also a data-transfer action that users should be warned about, especially in enterprise or sensitive environments.

External Transmission

Medium
Category
Data Exfiltration
Content
source "${SCRIPT_DIR}/config.sh"

# Get token from Feishu API
response=$(curl -s -X POST "${FEISHU_AUTH_ENDPOINT}" \
  -H "Content-Type: application/json" \
  -d "{
    \"app_id\": \"${FEISHU_APP_ID}\",
Confidence
70% confidence
Finding
curl -s -X POST "${FEISHU_AUTH_ENDPOINT}" \ -H "Content-Type: application/json" \ -d

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal