agent-Andri 安全扫描报告 — 低风险 | ClawSafe

10 /100

agent-Andri

Agent Andri — periodic status reporter writing to meeting-room file

A simple local status-reporting agent that periodically writes to a flat text file in a designated directory, with behavior fully aligned with its documentation.

技能名称agent-Andri

分析耗时20.6s

引擎pi

✓

可以安装

Skill is safe to use. The infinite while-loop in status_report.sh is a minor design concern for long-running sessions but does not constitute a security threat.

安全发现 2 项

严重性	安全发现	位置
低危	Writes to hidden dot-directory under $HOME 敏感访问 The script writes to $HOME/.openclaw/workspace/skills/meeting-room/to_leader.txt. This path is under a hidden dot-directory, which is benign but worth noting. `BASE_DIR="$HOME/.openclaw/workspace/skills/meeting-room"` → This is declared in SKILL.md and is the intended behavior. No action required.	`scripts/status_report.sh:9`
低危	Infinite loop without signal handling 敏感访问 The while-true loop in status_report.sh runs indefinitely with no trap for SIGTERM or SIGINT, which may leave zombie processes. `while true; do ... sleep 30; done` → Consider adding a trap to handle termination signals gracefully.	`scripts/status_report.sh:12`

资源类型	声明权限	推断权限	状态	证据
文件系统	`READ`	`WRITE`	✓ 一致	SKILL.md declares writing to ~/.../to_leader.txt; scripts/status_report.sh line …

目录结构

2 文件 · 983 B · 26 行

Shell 1f · 15L Markdown 1f · 11L

├─ ▾ 📁 scripts

│ └─ 🔧 status_report.sh Shell 15L · 386 B

└─ 📝 SKILL.md Markdown 11L · 597 B

安全亮点

✓ All capabilities declared in SKILL.md are used exactly as documented

✓ No external network requests or data exfiltration

✓ No credential harvesting or environment variable exfiltration

✓ No obfuscation, base64 encoding, or suspicious code patterns

✓ No sensitive file paths (~/.ssh, ~/.aws, .env) accessed

✓ No remote script execution (curl|bash, wget|sh)

✓ No supply-chain risks — no dependencies declared or used

✓ Script uses set -euo pipefail for safe shell execution