中文 EN

Scan Report

Scan New Files

⚠ Low Risk — Risk Score 25/100
Last scan:1 day ago Rescan
25 /100
BountyClaw Agent Skill
安全注册和操作龙虾众包平台Agent，使用bind_token机制绑定人账号
BountyClaw Agent registration skill with documented remote code execution capability; minor hardcoded credential placeholder flagged but no active exfiltration observed.
Skill NameBountyClaw Agent Skill
Duration28.6s
Enginepi
✓
Safe to install
Remove hardcoded password placeholder from example.js:191. Review remote code execution workflow in isolated environment before production use.

Findings 3 items

Severity Finding Location
Medium
Environment variable access not declared Doc Mismatch
Code reads BOUNTYCLAW_TOKEN from process.env but metadata.json does not declare environment:READ permission
const TOKEN = process.env.BOUNTYCLAW_TOKEN;
→ Add 'environment:READ' to permissions array in metadata.json or document this access in SKILL.md
 example.js:58
Low
Hardcoded password placeholder in code Sensitive Access
Pre-scan flagged 'your-secure-password' at line 191. While a placeholder, hardcoding credential examples in source code is poor practice
password: 'your-secure-password',  // Agent自身密码
→ Move credential examples to separate config/env files, use clearly marked placeholder values
 example.js:191
Low
Remote code execution capability with warnings Supply Chain
Skill is designed to download and execute arbitrary code from remote server. Warnings are present but risk remains
⚠️ 认领任务后会获得 download_url，但不要自动下载执行！
→ Ensure execution always happens in sandboxed environment; consider code signing verification for downloaded packages
 skill.md:72
ResourceDeclaredInferredStatusEvidence
Filesystem NONE NONE — No file operations in example.js
Network READ READ ✓ Aligned fetch() calls to puckg.xyz:8444 documented in SKILL.md
Shell NONE NONE — No subprocess or shell execution in code
Environment NONE READ ✗ Violation example.js:58 reads process.env.BOUNTYCLAW_TOKEN without declaration
1 High 5 findings
🔑
High API Key 疑似硬编码凭证
password: 'your-secure-password'
example.js:191
🔗
Medium External URL 外部 URL
https://www.puckg.xyz:8444
.env.example.md:8
🔗
Medium External URL 外部 URL
https://www.puckg.xyz:8444/api
example.js:14
🔗
Medium External URL 外部 URL
https://www.puckg.xyz:8444/api/agent/bind-token
skill.md:100
🔗
Medium External URL 外部 URL
https://www.puckg.xyz:8444/api/agent/tasks/TASK_ID/evidence
skill.md:312

File Tree

5 files · 24.7 KB · 919 lines
Markdown 2f · 629L JavaScript 1f · 231L JSON 2f · 59L
├─ 🔑 .env.example.md ⚠ Markdown 26L · 906 B
├─ 🔑 config.json ⚠ JSON 8L · 179 B
├─ 📜 example.js JavaScript 231L · 6.5 KB
├─ 📋 metadata.json JSON 51L · 1.8 KB
└─ 📝 skill.md Markdown 603L · 15.3 KB

Security Positives

✓ No credential exfiltration - tokens stay local and are not sent to third parties
✓ Proper JWT authentication flow documented
✓ Security warnings present for dangerous operations (remote code execution)
✓ No base64 encoding or obfuscation detected
✓ No subprocess/shell execution in the skill code itself
✓ No access to sensitive paths (~/.ssh, ~/.aws, .env files)