低风险 — 风险评分 15/100
上次扫描:2 天前 重新扫描
15 /100
agentwallet
EVM wallet tool for autonomous agents with built-in accountability. Creates, signs, and broadcasts ETH and ERC20 transfers on any EVM-compatible chain, then appends every transaction to a JSON log file.
Legitimate EVM wallet tool for autonomous agents with clear documentation. All capabilities are declared, no hidden functionality, and no malicious behavior observed.
技能名称agentwallet
分析耗时33.4s
引擎pi
可以安装
Safe to use when wallet keys and output paths are explicitly provided by the human. Always set --max-amount for x402 flows. Version pinning recommended for dependencies.

安全发现 2 项

严重性 安全发现 位置
低危
Unpinned dependency versions
requirements.txt specifies 'eth-account>=0.11.0' and 'requests>=2.28.0' without upper bounds or exact versions, potentially allowing supply chain risks from future breaking changes.
eth-account>=0.11.0
→ Pin to exact versions or narrow version ranges: eth-account==0.11.x, requests==2.31.x
requirements.txt:1
低危
Network WRITE capability not explicitly declared
Scripts make POST requests to external RPC endpoints and x402 APIs. While necessary for blockchain operations, this WRITE-level network access is not documented in SKILL.md.
RPC endpoints and x402-gated APIs
→ Document that the skill makes outbound HTTP POST requests to user-provided RPC URLs and x402 API endpoints.
SKILL.md:1
资源类型声明权限推断权限状态证据
文件系统 WRITE WRITE ✓ 一致 SKILL.md line 20-25, reads wallet JSON, writes agentwallet.json log
网络访问 READ WRITE ✓ 一致 Makes POST requests to RPC endpoints and x402 APIs (necessary for blockchain ope…
命令执行 NONE NONE No subprocess usage found in scripts
环境变量 NONE NONE No environment variable access observed
技能调用 NONE NONE No cross-skill invocation
剪贴板 NONE NONE No clipboard access
浏览器 NONE NONE No browser automation
数据库 NONE NONE No database access
18 项发现
🔗
中危 外部 URL 外部 URL
https://clawhub.com/skills/cdnsoft-wallet
SKILL.md:14
🔗
中危 外部 URL 外部 URL
https://rpc.linea.build
SKILL.md:27
🔗
中危 外部 URL 外部 URL
https://actors.dev/emails
SKILL.md:36
🔗
中危 外部 URL 外部 URL
https://mainnet.base.org
SKILL.md:38
💰
中危 钱包地址 加密货币钱包地址
0x3604712bd95ba2ff36b624f3ffeb6b73b34604ea
SKILL.md:44
💰
中危 钱包地址 加密货币钱包地址
0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913
SKILL.md:54
🔗
中危 外部 URL 外部 URL
https://cdnsoft.github.io/agentwallet
SKILL.md:139
🔗
中危 外部 URL 外部 URL
https://x402.org
index.html:361
🔗
中危 外部 URL 外部 URL
https://api.gateskip.org/solve/funcaptcha
index.html:377
💰
中危 钱包地址 加密货币钱包地址
0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48
index.html:413
💰
中危 钱包地址 加密货币钱包地址
0xfde4C96c8593536E31F229EA8f37b2ADa2699bb2
index.html:414
💰
中危 钱包地址 加密货币钱包地址
0xdAC17F958D2ee523a2206206994597C13D831ec7
index.html:415
💰
中危 钱包地址 加密货币钱包地址
0x50c5725949A6F0c72E6C4a641F24049A917DB0Cb
index.html:416
💰
中危 钱包地址 加密货币钱包地址
0x6B175474E89094C44Da98b954EedeAC495271d0F
index.html:417
🔗
中危 外部 URL 外部 URL
https://cypher.cdnsoft.net
index.html:422
💰
中危 钱包地址 加密货币钱包地址
0x4200000000000000000000000000000000000006
scripts/log_transaction.py:285
💰
中危 钱包地址 加密货币钱包地址
0x2626664c2603336E57B271c5C0b26F421741e481
scripts/log_transaction.py:286
📧
提示 邮箱 邮箱地址
[email protected]
SKILL.md:42

目录结构

5 文件 · 52.8 KB · 1308 行
Python 2f · 728L HTML 1f · 439L Markdown 1f · 139L Text 1f · 2L
├─ 📁 scripts
│ ├─ 🐍 log_transaction.py Python 409L · 15.8 KB
│ └─ 🐍 x402_request.py Python 319L · 11.9 KB
├─ 📄 index.html HTML 439L · 19.4 KB
├─ 📄 requirements.txt Text 2L · 37 B
└─ 📝 SKILL.md Markdown 139L · 5.7 KB

依赖分析 2 项

包名版本来源已知漏洞备注
eth-account >=0.11.0 pip Version not pinned
requests >=2.28.0 pip Version not pinned

安全亮点

✓ All functionality clearly documented in SKILL.md with accurate descriptions
✓ No subprocess or shell execution - uses Python requests library only
✓ No credential exfiltration - private keys used only for local transaction signing
✓ Safety checks implemented (--max-amount, --pay-to validation in x402 flows)
✓ No base64-encoded payloads or obfuscated code
✓ Cryptographically appropriate: uses os.urandom for nonces, standard EIP-712 signing
✓ Audit logging to append-only JSON is a positive security feature
✓ Clear warning about chmod 600 for wallet files
✓ No hidden HTML comments or suspicious instructions