ClawHub
Back to skill

Security audit

Teamgram Server Architecture

Security checks across malware telemetry and agentic risk

Overview

This is a documentation-only Teamgram architecture and deployment guide, but some sample deployment credentials must not be reused in production.

Treat this skill as a reference, not a production-ready deployment manifest. Before using its examples, replace all sample passwords, remove weak defaults, avoid root database credentials where possible, pin container image versions, restrict exposed ports, secure backups, and review the final configuration against official Teamgram and infrastructure security guidance.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Credential Access

High
Category
Privilege Escalation
Content
mysql:
    image: mysql:8.0
    environment:
      # ⚠️ 生产环境：使用强密码，通过 .env 文件或 secrets 注入
      MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PASSWORD:-changeme}
      MYSQL_DATABASE: teamgram
    volumes:
Confidence
95% confidence
Finding
The deployment example permits a dangerous default credential via `MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PASSWORD:-changeme}`, which can leave production systems running with a trivial root password if the environment variable is unset. In this context, the same guide also exposes multiple services and includes other weak defaults, making insecure copy-paste deployment more likely and increasing the chance of database compromise.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.